btSl" ■ •••.-• :=— ^ ••-,*'. .V,.-. ' I%^\ IN MEMORIAM FLORIAN CAJORI ♦ • AN ELEMENTARY TREATISE ON CURVES, FUNCTIONS, AND FORCES. VOLUME SECOND; CONTAINING CALCULUS OF IMAGINARY QUANTITIES, RESIDUAL CALCULUS, AND INTEGRAL CALCULUS. By benjamin PEIRCE, A. M. Perkins Professor of Astronomy and Mathematics in Harvard University. BOSTON: JAMES MUNROE AND COMPAJMY 1846. Entered according to Act of Congress, in the year 1846, by James Munroe and Company, in tho Clerk's Office of the District Court of the District of Massachusetts BOSTON: PRINTED BY THURSTON, TORRY & CO. 31 Devonshire Street. CONTENTS. BOOK III. CALCULUS OF IMAGINARY QUANTITIES. CHAPTER T. MODULUS AND ARGUMENT ...... CHAPTER n. IMAGINARY INFINITESIMALS CHAPTER m. IMAGINARY ROOTS OF EQUATIONS ..... CHAPTER IV. IMAGINARY EXPONENTIAL AND LOGARITHMIC FUNCTIONS 22 CHAPTER V. IMAGINARY CIRCULAR FUNCTIONS ..... 25 CHAPTER VI. REAL ROOTS OF NUMERICAL EQUATIONS .... 31 BOOK IV. RESIDUAL CALCULUS. CHAPTER L RESIDUATION iV!:?.Of>5^:57 3 12 13 43 IV CONTENTS. CHAPTER II. DEVELOPMENT OF FUNCTIONS, WHICH HAVE INFINITE VALUES 52 BOOK V. INTEGRAL CALCULUS. CHAPTER I. INTEGRATION ........ 63 CHAPTER II. INTEGRATION OF RATIONAL FUNCTIONS . . • . 69 CHAPTER III. INTEGRATION OF IRRATIONAL FUNCTIONS ... 77 CHAPTER IV. INTEGRATION OF LOGARITHMIC FUNCTIONS . . . 101 CHAPTER V. INTEGRATION OF CIRCULAR FUNCTIONS . . . .116 CHAPTER VI. RECTIFICATION OF CURVES ...... 124 CHAPTER VII. QUADRATURE OF SURFACES 166 CHAPTER Vm. THE CURVATURE OF SURFACES 185 CHAPTER IX. THE CUBATURE OF SOLIDS ...... 197 CHAPTER X. INTEGRATION OF LINEAR DIFFERENTIAL EQUATIONS . 207 CONTENTS. T CHAPTER XL INTEGRATION OF DIFFERENTIAL EQUATIONS OF THE FIRST ORDER ......... 239 CHAPTER Xn. INTEGRATION OF DIFFERENTIAL EQUATIONS OF THE SEC- OND ORDER ........ 274 CHAPTER Xni. PARTICULAR SOLUTIONS OF DIFFERENTIAL EQUATIONS . 287 CORRECTIONS, Page line 25 12 26 it t( (107) (110) (111) (114) 29 (138) (139) (140) 32 (149) <( (151) 48 7 53 23 54 14 55 (217) 64 (247) 65 (252) 66 2 I 89 (373) 90 (380) 92 1 t< 2 95 14 106 (439) (443) 107 (19 and 20) ) for read n 7C (n-\-fi) 71. 2B B. f.(a-\-i) dcf.(a + i) t I f(a-^i) dcf.(a — i) t I b^—2ab — a2 b^—3ab. finite infinite. 31 33. (^ + a) f.(x + a). /• ^0 f' ^i — /• ^0 • 3 4 m + l-\-np — ns m+l+n/j — ns The fraction is to be multiplied by (1 z^ + — z^.) "' 4 a 4a2 ' s n s pP—i bP—^. bx^ b x^ a a 10 ^10 biomial binomial. I 2A + 1 2A-1. 108 (458) The first and second members are to be divided by ^. 113 (501) 71 + 1 - n — 1. 114 (504) n + 1 n — 1. Vlll CORRECTIONS. Pagt line for read 119 (543) —— + — . 121 5 1 i 125 9 e2z^g-2, e2*_2^e-2*. 126 18 or F' T or F' T. 127 —1 P' p, " 11 Add^ or are supplements of each other (fig. 2). 128 2 and 3 pi p. 129 (574) e e". 134 (612) t V. (618) ± d=^. 136 11 are is. 137 J 3 (635) (630). 138 (643) fl" and y'/ ^^ and i^^ . " (648) f- fl 139 (649) If" if^. " (652) f- jfl 140 (658) (618) (658). " 9 fig. 5 fig. 6. "19 X V- 143 (680) 2e 4 e. " 13 and 15 F' F". 148 2 /;p /;. 149 2 tangent perpendicular. m/«Qo\
6.] MODULUS AND ARGUMENT.
Argument of any real quantity.
and the value of (1) becomes
R (cos. 6 + sin. ^. v' — 1). (8)
4. Corollary. Since two angles, which differ by two right
angles, have the same tangent, there are two values of fl less
than four right angles, which satisfy (5) ; and of these two
values, that one is to be selected which agrees, in the signs
of its sine and cosine, with (6) and (7). Any angle, which
differs from the value of & thus found by four right angles, or
by any multiple of four right angles, may also be taken as a
value of <5. Thus, if q^ is this least positive value of 6, the
general value of 6 is
5 = ^0 ± 2 n.^, (9)
in which n is any integer, and re is the ratio of the circumfer-
ence to the diameter.
5. Corollary. When the imaginary part of (1) vanishes, we
have
.B := 0, sin. ^ = 0; (10)
so that ^Q =: 0, a=:d=2w7r, COS. ^ = 1 ; (11)
or &o^n, 3 = zfc (2 W-(- 1) TT, COS. 5 =: — 1. (12)
and (11) corresponds to the case of a real positive
quantity J (12) to that of a real negative quantity.
6. Corollary. When the real term of (1) vanishes,
we have
^ = 0, COS. 6 = 0, ^0 = J TT or = J TT, (13)
whence
fl = zhj^±2n^ = ±(2 7i±})^, sin,5 = ±1. (14)
IMAGINARY Q,UANT1T1ES. [b. III. CH. I.
Equal imaginary quantities.
7. Theorem. When the quantity represented by (1)
vanishes, the real and the imaginary part, and the
moduhis, are each equal to zero, while the argument is
indeterminate.
Proof. For if
^4-jBV — 1 = 0, (15)
we have ^ = — JB \/ — 1 ; (16)
that is, a real quantity equal to an imaginary one, which is
impossible, and (16) cannot be satisfied, unless we have
^ = 0, jB^O; (17)
whence, by (4), R — Q, (18)
and, by (5), & is indeterminate.
8. Theorem. When hvo imaginary quantities are
equals their real and imaginary parts are separately
equal, and they have the same moduhis and argument.
Proof. For the equation
A+B^— l = A' + BW— ^, (19)
gives, by transposition,
A — A' -[- {B — B') s^ — \ = ^. (20)
Hence, by the preceding theorem, «
A — A' = (), B — B'=:Oy
or, A=:A', B = B; (21)
whence, by (4 and 5),
R = R', & = &'. (22)
<^ 11.] MODULUS AND ARGUMENT. 7
Conjugate quantities. Imaginary product.
9. Two imaginary quantities are conjugate to each
other, when they have the same modulus, and when
their arguments only differ in being of contrary signs.
Thus the conjugate of (8) is
R [cos. (— 6) 4- sin. (— a) . ^ _ 1] ; (23)
or, by trigonometry,
i? (cos. ^ — sin. 5.v/— 1). (24)
10. Corollary. Two imaginary quantities, which are
conjugate to each other, differ only in the sign which
precedes the imaginary part.
Thus A 4- Bs^ — 1 and A —Bs/— 1 are, by (8 and
24), conjugate to each other.
11. Theorem. The modulus of the product of sev-
eral imaginary quantities is equal to the product of the
moduli of the factors, and the argument of the product
is equal to the sum of the arguments of the factors.
Proof, a. When there are two factors
R (cos. & + sin. a. V — 1) and 7J'(cos. 6'4-sin.5'.\/ — 1), (25)
the product is
R R' [cos. 6 cos. &' — sin. q sin. S']
+ (sin. & COS. 6' 4" ^^^- ^' COS. &) \/ — 1, (26)
which, by (26 and 28 of Trig,), becomes
R R [cos. (^ + tv) -f sin. {d + 6') . V— 1] ; (27)
so that its modulus is the product of the two moduli, and its
argument is the sum of the two arguments.
8 IMAGINARY QUANTITIES. [b. III. CH. I.
Imaginary power.
h. A third factor miglit be multiplied by (27) in the same
way, that is, by multiplying its modulus by the new modulus,
and adding to its argument the new argument ; and this pro-
cess might be extended to any number of factors.
12. Corollary. If the factors are all equal, the pro-
duct becomes a power ; whence the modulus of a posi-
tive integral power of an imaginary quantity is the
same power of its modulus, and the argument of the
power is the product of its argument hy the exponent of
the power.
Thus
[/2(cos.6-|-sin.aV— l)]'^=-K"(cos.W5-fsin.w5.V— 1)-(28)
13. Corollary. When jR = 1, (29)
(28) becomes
(cos. 6 + sin. a. s/ — 1)" i:r (cos. n &-\- sin. n 6. y' — 1). (30)
Reversing the sign of ^
(cos.a — sin.a.V — l)'^ = (cos. n a — sm.n^.^/ — 1). (31)
14. Corollary, Half the sum of (30 and 31) is
cos.n a= J(cos.a-|-sin.a V— -1 )" + J(cos. 5 — sin.a. /y/— 1 )n. (32)
Half the difference of (30 and 31) is
sin. w a. y' — 1 = J (cos. a. -|- sin. ^. \f — 1)"
— J (cos. d —sin. 5. V — 1)" (33)
<§) 18.] MODULUS AND ARGUMENT. 9
Product of two conjugate factors. Imaginary quotient.
15. Corollary. By development, (32) becomes
nln — 1 ) _ . ^
cos.w<5 = COS." (3 :j — ^ — - COS.™ — ~(5 sin. 2 d
, 7i(n — l)(w — 2)(w — 3) . . ^
_[_ _i_^ — 11 TT^-i cos."-4 & sin.* d — &c. (34)
By developing and dividing by \/ — 1, (33) becomes
sin. w ^ zz: 71 COS."— ^ ^ sin. ^
n(7^_l) (n — 2)
1 . 2 . 3
cos."-3(3 sin.3(3 -}-&c. (35)
16. Corollary. The reverse of ^^^ 12 is, that the
modulus of a positive integral root of an imaginary
quantity is the same root of its modulus, and the argu-
ment of the root is the quotient of its argument divided
by the exponent of the root ; that is, since roots are
fractional powers, the rule of § 12 extends to the case of
positive fractional powers.
17. Corollary. The product of two conjugate factors
is equal to the square of the modulus.
For, in this case, (23 and 27) give
6 + ^' = ^ — a=zO, RR'=R2. (36)
18. Corollary. The reverse of ^ 11 is, that the mo-
dulus of a quotient is equal to the quotient of the Tnodu-
lus of the dividend divided by that of the divisor ^ and
the argument of the quotient is equal to the argument
of the dividend diminished by that of the divisor.
10 IMAGINARY QUANTITIES. [b. III. CH. I.
Imaginary power.
Thus
R'{cosJ'-}-smJ'.\/—\)
RicosJ-{-s'in.6.\/—l)
= ^'[cos.(^'— ^) + sin.(5' — ^)V— 1]. (37)
19. Corollauj. When ^' = 0, and R' — \ \ (38)
(37) becomes
z=-i [cos.(-^) + sin.(-a)V-l],
jR(cosJ + sin.5V— 1 ^
or
[i2(cos.^+sin.^.V— l)]-^=-R"^[cos.(— ^)+sm.(— 5)V— 1]
— i2-i(cos.5 — sin.^V— 1); (39)
and raising to the wth power, by means of (30),
[jR(cos.^4-sin.^V— l)]~"=^~"[cos.(-n^)+sin.(-w5)V— 1]
= K-™(cos.w^— sin.w^^/ — 1); (40)
that is, the rule of <§> 12 may he extended to the case of
negative powers.
20. Corollary. The rule of§ 12 may, then, be ex-
tended, by § 1, to all powers, real or imaginary.
21. Problem. To find the modulus and argumerrt of
the sum or difference of several imaginary quantities.
Solution. Let the given sum or difference be
r(cos.^4-sin.^V— l)i:r'(cos.^'+sin.^'.\/— 1)± &c., (41)
and let R be its modulus, and © its argument; we have by
(4 and 5) and by (9 and 29 of Trig.)
<5> 23.] MODULUS AND ARGUMENT. 11
Imaginary sura or difterence.
122= (r COS. ^±r'cos. a'i &c.)2-j- (r sin. ^ ± r' sin. 6'zh &c.)2
_ ^2 _|. r'2 + &,c. zb 2 rr' COS. (^ — ^') ± &c., (42)
r sin. 6 rt r' sin. d' r±z ^c. . ^ ^
tan. =: — ; T— - — J — (43)
r cos. 6 i r cos. 5' i &/C. '
22. Corollary. Since every cosine is less than unity, (42)
gives R^ < r2 -|- r'^ -|- &,c. -\-2rr'-\- &c.,
or i22 < (;. _|_ ,./ ^ &c.)2,
*or JR < r +r' + &,c.; (44)
that is, the modulus of the sum or difference of several
imaginary quantities is less than the sum of their
moduli.
23. Corollary. When there are only two terms in (41),
(42) becomes
2J2 -_ ^2 _j. y./2 _t- 2 r r' cos. (a — ^') ; (45)
and, therefore, R^ > r^ -{- r'^ —2r r',
or R >r — r'; (46)
that is, the modulus of the sum or difference of two
imaginary quantities is greater than the difference of
their moduli.
12 IMAGINARY QUANTITIES. [b. III. CH. II.
Imaginary infinitesimal.
CHAPTER II.
IMAGINARY INFINITESIMALS.
24. An imaginary infinitesimal is an imaginary quan-
tity, whose modulus is an infinitesimal.
The order of an im^aginary infinitesimal is the same
with that of its modulus.
25. Corollary. It follows from Chapter II. of the
Differential Calculus, and the preceding Chapter, that
all the propositions, lohich have hitherto been investi-
gated respecting real infinitesimals^ may he extended to
imaginary infinitesimals.
Ǥ> 27.] ROOTS OF E(iUATIONS. 13
Roots of a binomial equation.
(48)
CHAPTER III.
IMAGINARY ROOTS OF EQ,TTATIONS.
26. Problem. To solve a binomial equation, and re-
duce all its imaginarij roots to the form of% 1.
Solution. Let the equation be
Ax''— M, (47)
in which A and 31 are real or imaginary, and a a positive
integer. When (47) is divided by A by means of ^ 18, it is
reduced to the form
X'^ -ZZZ 777,
in which m is of the form of ^ 1. Let then
m z=z r (cos. 6 -\- sin. 6. \/ — 1), (49)
or x« = r (cos. 6 + s'"- ^- V — !)• (50)
The ath root of (50) is, by § 16,
^ ^ A
x=z/v/r. (cos. {-sin. — .\/ — 1). (51)
a a ' ^ '
27. Scliolium. Since has, by (9), an infinity of values,
(51) would at first sight appear to have a like infinity of values.
But, by (9),
& fl„ 2 7Zrr
— = — dt , (52)
a a a ^ '
14 IMAGINARY QUANTITIES. [b. III. CH. III.
Number of roots of a binomial equation.
whence the values of x are identical, when they correspond to
values of <^, for which the difference of the values of n is equal
to a, or is some multiple of a. Now, by subtracting from any
value of n the greatest multiple of a contained in it, a remain-
der is obtained, which is less than a. The number of differ-
ent values of x is, therefore, the same with the number of posi-
tive integers (zero included) which are less than a ; that is, the
number of values of x or the number of roots of equation (48)
is just equal to a.
28. Corollary. When m is real and positivCj (11)
gives
«, / 2n-n: . 2nTv \
X z= \/m i cos. zb sin. — — .V — 1 I > (5^)
in which the double sign renders it unnecessary to no-
tice those values of n which exceed the half of a.
29. Corollary. The value of n
n = 0, (54)
reduces (53) to its real positive root
a
X z=z s/ m, (55)
30. Corollary. When a is even in (53), the value of n
nr=.\a, ' (56)
2w^ .
gives = ^> V^'y
a
X =. — \/m. (58)
<5> 33.] ROOTS OF Eq,UATIONS. 15
Every equation has a root.
31. Corollary. When in is real and negative, (12)
gives
«^ / 2/1 + 1 . 2n + l , A ,^^^
zi=V — wifcos. TTitsin. >T.\/ — II, (o9)
in which the double sign renders it unnecessary to no-
tice those vaUies of n which exceed the half of a.
32. Corollary. When a is odd, the value of n
—1
n = %^. (60)
2n + l ' ,^,,
gives Tc ^::^ n^ (61 )
X z=: — \/ — m. (62)
33. Theorem. Every equation has at least one real
root or one imaginary root of the form (1).
Proof. Let all the terms of the equation be transposed to
its first member, which reduces it to the form
/.x = 0. (63)
Let now x^ be any real or imaginary value of x, for which
the value of this first member neither vanishes, nor is infinite,
and let h be an infinitesimal ; let also dl.f.x^ be the first dif-
ferential coefficient o{ f.x^ which does not vanish ; and (533
of Vol. I.) gives
f{x^+h) =r/.x, + ^^ 3 ^^ d:.f.x^' (64)
16 IMAGINARY QUANTITIES. [b. III. CH. IIT.
Equations whicli have finite roots.
Again, let i be «in assumed real infinitesimal, and let h be
determined to satisfy the assumed binomial equation
r^r — - '^"J.^o = - if^x^. (05)
This value of h, being substituted in (G4), gives
/.(/, +70 =/.x, - z/:r, = (l-O/.^o ; (66)
so that if r is the modulus of/.a-Q, that of/.(.TQ + /O ^^> ^7
§11, (1 — i)r, and therefore less than that o^ f.x^. The least
possible modulus of y.a; is then less than r, unless r is zero ;
this least modulus must then be zero, and the corresponding
value of X is a root of the equation (63).
34. Scholiiun. The preceding argument does not
exclude infinity from being the root of the given equa-
tion, so that the following is a convenient statement of
the above theorem ;
Every equation has at least one finite root of the form
(1), wheiiy after it is reduced to the form (63), it does
not vanish for an infinite value of tlie variable.
35. Corollary. If the first member of (63) is a polynomial
of the form
xn -]_ a 2-"-i -\- h 3;"-2 + &c., (67)
and if x' is a root of the equation, this polynomial must be
divisible by x — x' ] and the quotient must be a polynomial of
the (/^ — l)st degree, which must be divisible by a similar
factor X — x", and so on.
<5) 37.] ROOTS OF EQUATIONS. 17
Tlie conjugate of a real function.
Hence (67) must he the continued product of n dif-
ferent factors of the form {.V — x^); that is, the equa-
tion
xnj^a x«-i + h 2;"-2 -f &,c. = (68)
must have n roots of the form (1), whether a, b, Sf'c. be
real or imaginary,
36. A real function is one, which has real values for
all real values of ihc variable, and has not imaginary
values, unless the variable is imaginary.
37. Theorem. The conjugate of a real function is the
same function of the conjugate of the variable ; or,
algebraically, if
P + QV-1=/.(P + 2V-1), (C9)
where /. denotes a real function, then
P-Qs/-\^f.{p-qs^-\). (70)
Proof. The function, which is the second member of (69),
may be developed and arranged according to powers of \/ — 1.
Let, then, the aggregate of all the terms which are independ-
ent of v' — 1) 'if'd of those which are multiplied by even
powers of \/ — 1 be denoted by P ; while the aggregate of
all those terms which are multiplied by odd powers of \/ — 1,
is denoted by Q'. The value of P is real, and remains un-
changed by changing v' — 1 ^^ — \/ — 1, while that of Q'
is reversed ; that is, the value of the function is changed
from P + Q' to P — Q'. (71)
2*
18 IMAGINARY QUANTITIES. [b. III. CH. III.
Every real equation has at least two roots.
But the quotient of Q' divided by \/ — 1, containing only
even powers of \/ — 1, is a real quantity, which may be de-
noted by Q, that is,
Q'=zQx/— 1, (72)
P+Q' = P+QV — 1; (73)
so that by reversing the sign of \/ — 1, (69) is changed to
(70).
38. Corollary, When Q = 0, (74)
(69 and 70) become
-P=/-(i' + 2V-i)=/.(i'-sV-i); (75)
that is, every real value of a real function corresponds
to two different values of the variable, which are con-
jugate to each other.
39. Corollary. When P = 0, (76)
(75) becomes
0=/. (p + 5\/-i)=/. (p-?Vi); (77)
that is, ivhen the function ^ which is the first meniher
of (63), is real^ tlie conjugate of every imaginary root
is also a root of the equation.
40. Corollary, If x' is a root of the equation (68), when
a, fc, &c. are real, and if x" is the conjugate of x' , x" is also
a root of this equation, and the first member is divisible by
the product
(z — X') (x — x") — x^— {x' + x'') X + x' x". (78)
<§. 40.] ROOTS OF EQ,UATIONS. 19
Number of real factors of a real polynomial.
If r is the modulus of z' and & its argument, (8, 24, and 36)
give
%' -\-x" — "ir COS. 6, x' x" = r2 ; (79)
whence (78) becomes the real factor
x2 — 2rx cos. 6 + r2 ; (80)
so that «???/ real polynomial of the form (67) is the con-
tinued product of as many real factors of the form x — x'
as the equation (68) has real roots, multiplied by the co7i-
tinued product of half as many real quadratic factors
of the form (80) as (68) has imaginary roots.
41. Examples.
1. Decompose z'^ — h"^ into a continued product of real
factors of the tirst and second degree.
Solution. The equation
a;7 — 67 _ 0, or x^ = b^ ,
gives in (48) m = b"^ , = 7;
whence (53) becomes
X z=z b (cos. f n TV :iz sin. f n re . a^ — 1 ) j
which becomes, by putting saccessively for Ji all integers less
than half of 7,
X z=z b,
X =1 b (cos. f 7r i sin. f tt . \/ — 1),
X z=z b (cos. f- ^ rt sin. f ^ . \/ — 1),
X z=: b (cos. f n zh sin. f ^ . \/ — 1 ) ;
20 IMAGINARY Q,UANT1TIES. [b. III. CH. III.
Decomposition of a function into real factors.
SO that, by (SO), the continued product is
X' —b' ={x-^b) {z2—2bx COS. f .T-f&2)
(x2_2 6 2;cos. f -f + 62) {x2—2bxcos.^^-\-b2).
2. Decompose x^ -f~ ^* ^^^^ ^ product of real factors of the
first and second decree.
Solution. The equation
x^ -{- b^ =1 0, or a-4 = —. 6*,
gives in (48)
?« = — b^, — ?n =z b^, a =z ^;
whence (59) becomes
x = b (cos. i (2 w -]- 1 ) TT i sin. ^ (2 71 + 1 ) rr . V— 1 ) J
which becomes, by putting successively for 7i all integers less
than 2,
X = b{c0S.irt-^sm.lrc,^—l) = b(h\/2:hiV^'V—l)y
a:=6(cos.f ^±sin.f:T.\/— 1) = ^(— W-=FiV-V— 1);
so that, by (80), the continued product is
3c4_|_54_(3;2_26a;cos.-i-7r-|-62)(x2_263;cos.j7r + 62)
_ (3;2 — ^ 2 . 6 X + 62) (3;2 _|_ ^2 . fe X + 62).
3. Decompose x^ — 6* into a continued product of real
factors of the first and second degree,
Ans. (x — 6) (x + 6) (x2 + 62).
4. Decompose x^ + 6^ into a continued product of real
factors of the first and second degree.
Ans. (x+6)(x2— 26xcos.|7i:+62)(x2_26a;cos.f;T + 62).
$ 40.] ROOTS OF EQ,UATIONS. 21
Decomposition into real factors.
5. Decompose x^ — h^ into a continued product of real
factors of the first and second decree.
Arts, {x—b) {x+b) (x2_|_5a;_j_52) (^x^ — bx + b^).
6. Decompose x^ -\- b^ into a continued product of real
factors of the first and second deorree.
Ans. (22-1-^3.6.2;+ 62) (a;2-)-62) {z2 — ^S.bx-{-b2).
22 IMAGINARY Q,UANTITIES. [b. III. CH. IV.
Imaginary power.
CHAPTER IV.
IMAGINARY EXPONENTIAL AND LOGARITHMIC FUNCTIONS.
42. Problem, To reduce an imaginary power of a
real quantity to the form (1).
Solution. Let the exponent of the power be ^-[- B a^ — 1,
and let R be the modulus and the argument of this power
of the real quantity a, that is, let
a-^+^v-i — jR(cos. + sin. 0.^—1). (81)
The infinitesimal power i of this equation is by (28)
«(^+5v-i) — Ri (cos. i + sin. io.^—l). (82)
Hence by (418 of Vol. I. and § 22 of Plane Trig.)
l-^i{A + B\/—l)\og.a—{l + i\og.R){l + i0^/—l)
— l+i{\og.R-{-0\/—l), (83)
Hence, by <5> 8, and using
e = the base of the Neperian logarithms, (84)
log. R — A log. a z= log. a-^, R — a-^ , (85)
B log. a z= = log. a^, aB[z= c©; (86)
which, substituted in (81), give
a^-^B^-i — a-A (cos. B log. a + sin. B log. a . \/—\). (87)
<§> 47.] LOGARITHMIC FUNCTIONS. 23
Imaginary logarithm.
43. CuroUarj/. When ^ == 0,
(87) becomes
a^v/-i — COS. B log. a + sin. B log. a . \/— 1. (88)
44. Corollanj. When a =: e,
(87 and 88) become
e-^+B^-i — e-^[cos.B + ^m.B,s/—\), (89)
gB./-i _ eos. B + sin. B . V— 1. (90)
45. Corollary. Reversing the sign of B, (89 and 90) be-
come
^A-B^f-i — e.^(cos. B — sin. S . \/— 1), (91)
e-5v-i z= cos. jB. — sin. ^. V— 1- (92)
46. Problem. To reduce the logarithm of an imagi-
nary quantity to the form (1).
Solution. Let r be the modulus and <3 the argument of the
imaginary quantity, and (90) gives
r{Q.os.&-\-s\n.&./^—\) = re(^^-^] (93)
the logarithm of which is
log. [r (cos. (3 -|- sin. 6 . /y/ — 1 )] =: log. r -\- log. c^ -^—"^
=:log. r + 6V— 1. (94)
47. Corollary. By (4, 5, and 94)
log.(^+i?V— l)=logV(^2_[_2J2)_|_tan.[-i]^V— 1
= J log. (^2 _j_S2 )_ptan.[-i]^. V— 1 ; (95)
24 IMAGINARY QUANTITIES. [b. III. CII. IV.
Number of the logarithms of a number.
and as there is an infinity of values of
a =1 tan.[-i] — ,
A
every quantity^ real or hnaginary^ has an infinity of
logarithms J of w J lich there is never more than one real
logaritlim^ and that^ hy ^ 5, only ivhen the quantity is
real and positive.
48. Corollary, By § 5, when A is positive, and
B — Q,
(95) becomes
log. A = log. A^'Znn s/— 1, (96)
in which log. A of the second member is the real value of this
logarithm.
49. Corollary. By § 5, when A is negative and
5 = 0,
(95) becomes
log. A — log. (_- ,4) ± (2 n + 1) ^ V— 1. (07)
50. Examples.
1. What is the logarithm of J\/2(l +\/— 1)?
2. What is the logarithm of \/3 + \/— 1 ?
Ans. log 2 + (^ zh 2 n) 71 ^— 1.
<§> 51.] CIRCULAR FUNCTIONS. 25
Sine and cosine of imaginary angles.
CHAPTER V.
IMAGINARY CIRCULAR FUNCTIONS.
51. Problem. To reduce the sine and cosine of an
imaginary angle to the form ( I).
Solution, a. Let the angle be B /s/ — 1, which being sub-
Btituted for B in (90 and 9^), gives
e-B — COS. B V— I + sin. B sf— 1 . V— 1, (98)
e^ = COS. 2J^/— 1 — sin .BV— 1 . V— 1- (99)
One half of the sum of (98 and 99) is
COS. Bs/—\ — l{c^-\- e-B), (100)
One half of the difference of (98 and 99), multiplied bj
iV — 1, is
sin. ^V— 1 = i{e^ — e-^)\/— 1. (101)
b. When the angle is A -{- B\/—}, (100 and 101) give
Bin,(^-f"^V' — l) = sin.^cos. B/v/ — 1 -f-cos. A s\n.B\/ — 1
= ^sin.4(e^-[-e-^) + icos..4(e« — e-^)>s/— I ; (102)
eos.( A-\-B\/ — I ) = cos. A cos. B\/ — I — sin. ^4 sin. By/ — 1
= icos.A(e^ + e-^) — is\iuA(e^—e-^)^^l. (103)
3
26 IMAGINARY QUANTITIES. [b. III. CH. V.
The imaginary angle, whose sine exceeds unity.
52. Problem. To reduce the imaginary angle, the
absolute value of whose sine is greater than unity, to
the form (I).
Solution. Let the given sine of the angle be db (1 + «),
and let the required angle he A -{- B \/ — 1 ; it is evident
from (102) that, when the sine of the angle is real,
cos.^(e^ — e-^') = 0; (104)
that is, either e^ = c-^, (105)
whence e^^ = 1, 2 B = 0, i5 = ; (106)
in which case the given angle is real, and the absolute value
of its sine cannot exceed unity ;
or cos. J = 0, A -=znn^ (107)
sin. yl = =b 1, (108)
whence, by (102 and 103), (109)
sin.(^ + /^V— l) = sin.(7i7r + SV— 1)
=:iM'^^ + ^-^) = i(l+«), (110)
COS.(^ + i^\/— l)=:C0S.(«7r4-i?V— 1)
^zp^(e^— e-^)x/— 1
_zp^(— .2a— «2)z==Fv/(2a+a2)V— 1. (HI)
The sum of (HO), and (111) multiplied by \/ — I, is
c^ =: 1 + « ± V (2 « + a^), (112)
whence
JB = log. [I +fl±\/(2fl + fl2)]
= i log. [1 + a + \/(2a + a^)l (113)
and the angle is
n^ d= log. [I + a+\/(2a + «2)].v'— 1. (114)
§ 54.] CIRCULAR FUNCTIONS. 27
Imaginary circular functions.
53. Examples.
1. Reduce tang. {A-\- B \/ — 1) to the form (I).
2 sin. 2^ (e2^ — 6-2^)./— 1
^'^^' ^5_|_e-25 +2 COS. 2 ^ "^ c2^ + e-2^ + 2 cos.24* ^^^^^
2. Reduce tang. S\/ — 1 to the form (1).
3. Reduce tang, f— ^] B a^/ — 1 to the form (1).
Ans. When B is absolutely less than unity, it is
±w^ + J[log.(l+^) — log. (1— J5)].V— 1. (117)
When B is positive and greater than 1, it is
±(n+J).^ + Hlog.(S+l)-log.(^-l)].x/-l. (118)
When B is negative and less than — 1, it is
± {n+ih+i[\og.-{l+B)-\og.(l-B)W-l' (119)
When S := =i= I, it is
^drOD.V— 1. (120)
54. Equations (100 and 101) have suggested a new form of
notation of great practical value, and for which tables have
been constructed, similar to the common trigonometric tables.
It consists in representing — \/ — I . sin. B \/ — 1 and
COS. B \/ — 1 by Sin,^ and Cos. ^, which only differ in
their initial capital letters from the common trigonometric no-
28 IMAGINARY QUANTITIES. [b. III. CH. T.
Potential functions.
tatioii ; this notation may also be extended to the other trigo-
nometric functions. These new functions are called potential
functions. We have, then,
fi- -r, , , . ^ # , sin. Bx/ — 1
= i{e^-e-^), (121)
Cos.B=icos,Bx/—l =i{eB + e-B)^ (122)
-„ „ Sin. 5 inner. Ba^ — 1
55. Corollary. The differentiation of (121 - 123) gives
d,. Sin. 5 = J (e^ + e-B) — Cos. B, (124)
rf,. Cos. J5 := 4 (e^ — e-^) — Sin. ^, (125)
«//ran. jB z= ^7^—, = 7T-V^ = Sec.2J5. (126)
' C0S.2^V— 1 C0S.25 ^ '
56. Examples.
Demonstrate the following equations.
1. C0S.2 B — Sin.2 B = 1. (127)
Solution. By (121 and 122)
C0S.2 B = 1 (c25 + 2 + C-2^)
Sin.2 JB =: ^ (e25 __ 2 _|_ g-sB)
Hence Cos.2 ^ _ sin.2 ^ — 1.
2. Sin. {BdtiB') = S\n.B Cos.B' ± Cos. B Sin. B' ( 128)
3. Cos.(JB ± JB') = Cos. B Cos.^'i Sin. B Sin. 5' (129)
4. Sin.(B + JB') + Sin. {B—B') = 2 Sin.£ Cos.^' (130)
^ 54.] CIRCULAR FUNCTIONS. 29
Potential functions.
5. Sin.{B + B') — Sm.{B—B') — 2Cos,BSm,B' (131)
6. Cos.{B-{-B') + Cos.(B—B') = '2Cos.BCos.B' (132)
7. Cos.{B+ B') — Co^B—B') = 2 Sin. ^ Sin.J3' (133)
Sin. ^ + Sin.^- _ Tang. Hg + ^0 f.^..
' Sin. ^ — ^m.B' ~ Tang. ^ (B — B) ^ '
Cos 7? Cos Tl'
^- Co:.B + Co:.B- = Tan.i(iJ+B')Tan.J(B-B') (135)
10. Sin. 2B z=2 Sin. B Cos. 5 (136)
11. Cos. 2 5 = Cos.2 J5 + Sin.25 (137)
z= 1 + 2 Sin.2 B
= 2 Cos.2 B — 1
12. Sin. ^jB=: x/[|(Cos. 2 5 — 1)] (138)
13. Cos.^B = ^[^{Cos.2B + l)] (139)
t4 rn . ^ , / Cos. 2 5 — 1 \
14. Ta„g.JB = v(c„i:2^+l) (140)
15. Ta„g.(iJ±B')=. 5^^:^-^^^, (141)
1 zh 1 ang jB 1 ang. B' ^ '
16. Tang. 2B= ^ . (142)
14- 1 ang. 2 5 ^ '
17. d^.SinS-^}z= {l + z^)-i = -^^±--. (143)
Solution. Let 2: == Sin. 5, or jB = Sin.[-i] z
Then by (124 and 127)
ef,.^.x = Cos.jB = -v/(l + Sin.2 5) = ^(i -j. ^2)
3*
30 IMAGINARY QUANTITIES. [b. III. CH. V.
Potential functions.
and by (Vol. I. 566)
d,. Sin.[-i] X := 4, B =
dcB'^ \/(l+^2)
18. rf..C03.[-i]x= (.^-lH:zz ^^J_^ - (144)
19. J,.Tang.[-i]x=-j-^. (145)
20. Sin. X = X + ^^-^ + --^^^^ + &c. (146)
21. Cos. X = 1 + -- + :^^^-^-^ + &c. (147)
§ 59.] REAL ROOTS. 31
Stern's method of solving numerical equations.
CHAPTER VI.
REAL ROOTS OF NUMERICAL EQUATIONS.
57. While the imaginary roots of equations are of
great subsidiary value in mathematical investigations,
and frequently admit of curious and interesting inter-
pretations in physical inquiries, real roots are the prima-
ry objects of attention, and methods of determining
their numerical values are exceedingly important in
practice. Ster7i^s method is the simplest which has yet
been published, and is of almost universal application.
58. If the values of a given function and of its suc-
cessive differential coefficients, as far as the ?ah, are
found for a given value a of the variable ; and if the
successive signs of these values are placed after each
other, the row of signs thus formed is, in this chapter,
called the nth row of signs (a), or simply the nth row
(a), or the row (a) ; any pair of successive signs in this
row is called a permanence^ when the signs are alike,
and a variation, when the signs are unlike.
59. Theorem. If a function and its differential co-
efficients inferior to the nVa all vanish, but the ?/th does
not vanish, for a value a of the variable, the nth. row
of signs (a + i)j * being an infinitesimal, consists
32 IMAGINARY QUANTITIES. [b. III. CH. VI.
Signs of vanishing functions.
— 1 ^ — -
wholly of permanences, while the nih row (a — i) con-
sists wholly of variations.
Proof. It follows from (Vol. I. 533), that if/, x is the giv-
en function
/• (« + = Y^ — ;;■ ''-■ J"-"- ('^^)
the differential coefficients of which, taken relatively to i, are
t.f. ia + = 1T2T3 r:Wl)- ''■^- " = "-^^^'^'
^ f(„A.i^- '•-' .. . (n-\)f.(a+i)
dl.f. (« + - i.2.3...(„_2)- ''-f-"- i '
&c. &c. &c. (149)
that is, all the terms of the series
/.(« + /), d..f.{a + i), dl.f.{a + i),&.c. (150)
have the same sign.
But the reversing of the sign of i in these equations gives
nf. (a—i)
dc'f- {a—i) =
i
ctl.f. {a — i) = —^ ^-^^ S&c. (151)
that is, the signs of any two successive terms in the series
/. (a — 0, d,.f.{a-~i), e/. (« — 0, &c. (J52)
are unlike, and the terms are alternately positive and negative.
60. Corolla?^. If, in a series of the successive dif-
ferential coefficients of a function terminating with the
<5» 61.] REAL ROOTS. 33
Number of real roots between given limits.
nth, all vanish except the 7i\h for a vahie a of the vari-
able, the signs of this series will in the row of signs
(a -}- i) constitute a series of permanences, and in the
row (a — i), a series of variations.
61. Theorem, If the first member of the equation
f.x = (153)
is coiithmons between the values a and b of the variable,
a being greater than b. if tJte nnniber of permanences
in the nth row of signs (a) exceeds tJie number of per-
onaneuces in the ntJi row (6), and if the excess is denot-
ed by V, the number of real roots of (153), ivhich are
included between a and b, cannot exceed v.
Proof. For while the value of x varies from a to b, a
change of sign can occur in the row of signs, only when f z,
or one of its differential coefficients, or a series of them, pass-
es through zero. Now, the case of a single function being
included in that of a series, when a series of these functions
vanishes, a number of permanences must, by ^^ 59 and 60,
be lost, equal to the number of functions. If, then, this series
begins with f x, as it must when the variable is equal to a
root of the equation, one permanence, at least, nmst be lost;
that is, there is a loss of one or more permanences in the row
of signs, corresponding to every real root of the equation.
If the vanishing series does not begin witli f. r, and con-
sists of an even number of functions, the sign of its first func-
tion is, by (148-152), the same with that of the function
which follows the series, both before and after vanishing. The
relation of the first sign of the series to the sign which pre-
34 IMAGINARY (QUANTITIES. [b. III. CH. VI.
Number of real roots of an equation.
cedes the series is, therefore, unchanged ; and the loss of per-
manmrcs is exactly equal to the even number of terms of the
vanishing series.
If the vanishintr series consists of an odd number of func-
tions, the sign of its first function is reversed when it vanishes.
If, therefore, it has, before it vanishes, the same sign with the
preceding function, another permanence is liere lost, which is
to be added to those before noticed. But if it has, before it
vanislies, the opposite sign to the preceding function, a new
permanence is introduced, when it vanishes, which is to be
subtracted from the number of the others. In one case, there-
fore, the ivhole number of lost permanences is one greater
than the odd number of terms in the vanishing series ; and,
in the other case, it is one less than this number.
In any case, the nuinber of lost permanences is, at least, as
great as the number of real roots of the equation.
62. Corollary. When the loss of permanences does
not arise from a real root of the equation, the number
of lost permanences is even ; so that if the number of
lost perrnanenres is odd, that is, ifv is odd^ the equation
must have at least one real root betioeen a and b.
63. Problem. To find all the real roots of an equa-
tion.
Solution . Reduce the equation to the form (153),
simplify it as much as possible; and determine, as
nearly as possible by inspection, those limits between
which the different real roots must be, if there are any.
Find the successive differential coefficients of the first
«5> 63.] REAL ROOTS. 35
Stern's" method of finding the real roots.
member^ until one is obtained which docs not vanish be-
tween two limits a and b, between which there may be
real roots. Let this be the nth differoUial coefficient.
If^ then., a being greater than b, the nvmber of per-
manences in the nth roio of signs (a) is the same ivith
that in the row (b), there is no real root between a and
b. If the difference between the number of 'permanen-
ces is even, the question of a real root between a and b
is undecided ; and if this difference is odd, there must
be such a root.
Let, then, the mth differential coefficient be the hi^'h-
cst one, of which the sign is different in tlie row (a) and
in the roio {b). The equatiofi
f/,.- + i/. xz=0, (154)
can then have no real root between a and b, luJtile the
equation
d^rf.x-^, (155)
must have one, which can be found by the process gicen
in the sequel of this solution. If c is the rnnt of (155),
it may also be a root of (153), which can be discovei^ed
by trial.
However this may be, the preceding process is to be
repeated for the limits a and c -\- i, i being an. infinites-
imal, and also for the limits c-\-i, and b, usitig the
mth rov) of signs instead of the nth. A continuation
of the process '//nist finally lead to a division of the lim-
its from a to b, into sets of limits so norroiv, that, be-
36 IMAGINARY QUANTITIES. [b. II I. CH. VI.
Stern's method of finding numerical roots.
tween each set there con only be one real root of ( 153)
and no real root of the equation
chfx = 0. (156)
Lei a' and h' he a set of these limits, and if they are
far apart, substitute for x, in the first member of[ 153),
different numbers, the various integers for instance^
between a' and b' , until one is found which does not dif-
fer nincfi from the required root, and denote this first
approximalio/i to tlie root by x^. Tlien, if the exact
root is XQ-\-h, ive have by (Vol. I. 532)
/. (..„+/,) =/.r„ + h ch.f. {r„+« A) =0, (157)
whence, by neglecting &h, the approximate value of h
is obtained, which is
h = -j^ (158)
and from the new approximation to the root x^ + A,
which is thus found, a neio approximation, can be ob-
tained ; and so on, to any required degree of accuracy.
64. Corollary. The rate of approximation can be readily
determined ; for if two successive values of h are h and A'
corresponding to a^ and x^, so that
rro z:z To + /i (159)
the error of x^ differs from h by a quantity much smaller than
A ; and that of x^ is nearly equal to h' . Now suppose
A<(Tvr (160)
<^ 65]. REAL ROOTS. 37
Rate of approximation.
and we have by (158) and by Taylor's theorem
f.x'^=f{x,+h):=f,x,+ d,.f,x^,h-\.ldi.f,x,\h^+ &c. (161)
c/,/.2:;=rt/,./.Xo+&c. (162)
but by (158)
f.x, + d,.f.x,.hz^O (163)
/.x; = J ./?./. 2:0. A2+&C. (164)
whence neglecting W^, &c.
If now we find
we have, neglecting the signs,
/*'<(tVP+* (167)
and therefore if one approximation is accurate to s places
of decimals, the next will be accurate ^0 2 s + A: places,
65. Corollary. Since the real root is exactly
^==^0 + ^1
we have ^0 = ^ — ^'» (168)
whence by (153 and Vol. I. 532)
/. x, = f(x — h) =/. X — h,d,,f {x^&h)
=z — hd,.f(x^6h), (169)
or neglecting & k
/. Xo= _ A d,.f. x={Xq^ x) d,.f. X. (170)
In the same way for another hypothesis x^, we have
/. z; r= _ h' d^.f X = (x; — X) d^.f X. (171)
nr
38 IMAGINARY QUANTITIES. [b. III. CH. T.
Rule of false or double position.
The difference of (170) and (171) is
/.7,-/..r; = (zo-r;)c/,./.x (172)
and the quotient of (171) by (172), is
f'" ^^'.^^I^ (173)
which is identical with the famous rule of false, or
rule of double positin??., iti arithmetic; and this admir-
able rule, the principle of which is obviously at the
foundation of all higher mathematics, and pervades all
practical science in some form or other, is sufficient for
obtaining, with ease and accuracy, the most important
numerical results.
66. Examples.
1. Solve the equation
xlog/ z— 100 —
in which log.' denotes the common tabular logarithms.
Solution. The theory of logarithms gives
log.' X z=z log.' e . log. X.
Hence if f.x=^x log.' x — 100
d,.f.x = log.'z + log.'c
The value of d^^'f- ^ is positive between the limits
X = 0, and x zn oo
and d^.f.x is negative between the limits
X = 0, and x =. c~~^
^66.]
REAL ROOTS.
39
Solution of numerical equations.
at both which limits /'. x is negative, and the given equation
has therefore no real roots between these limits. But d^.f.z
is positive between the limits x=:c~' and x = 0, at which
limits f, X has opposite signs, and the given equation has,
therefore, only one real root, which is between these limits.
A very few trials show, then, that the root is not far from
60, for which value
f.x — 1, d\.f.x— 1-78 + 0-43:= 2-21
<^,./.xr= -0072, A: = 2
and the rest of the calculation may be arranged as in the fol-
lowing form, in the first column of which are placed the suc-
cessive values of y.x, in the second those of rf^'/^ ^» sind in
the third those of x.
7.
0084860
221
219017
60
57
569612
2. Solve the equation
X — cos. X =: 0.
Ans. 56-9612.
Ans. 0-7391.
3. Solve the equation
X — tang, X = 0.
Ans. There are an infinity of roots, one being contained
between each set of limits
n n and (^ + J) tt
in which any integer may be substituted for 7i, the value be-
tween ^ and J ^ is 4*4934.
^•'•f
BOOK IV.
RESIDUAL CALCULUS.
4*
A
BOOK IV.
RESIDUAL CALCULUS.
CHAPTER I.
RESIDUATION. "S^^'
1. For every finite value of x, which satisfies the equation
/x = 00 , that is, ~ = (/.x)-i = 0, (174)
/• ^
the first term of Taylor's theorem (Vol. I. 442) is infinite, and
the development of y. (x ~\- h) by that theorem is impossible.
In this case, if i is an infinitesimal, f. (x -\- i) is infinite ;
and if we suppose it to be of the ?wth order of infinity, the
expression
imf,(x + i) (175)
is of the order zero, and is usually finite, as in § 26 of the
Differential Calculus. The quantity
h^f.(x + h) (176)
may then be developed, by MacLaurin's Theorem (445, Vol. I.),
as a function of A, and the result will be of ihe orm
h'"f,(x-\-h) ==A -{- Bh-\-&.c.
_j« Qh^-^-{.Rhm-i 4- Sh^^ + r/i'^+i + ifcc. (177)
44 RESIDUAL CALCULUS. [b. IV. CH. I.
Residual. To residuate.
which, divided by 7i'", gives
+ Q A-2 4- 7? A-i 4- >S 4- Th + &c. (178)
that is, f.{x -\- h) can, evcti for a value of x which
satisfies (174), he developed in a series consisting of
two pa; ^5, one of luliich
S -\- Th ^ 6cc. (179)
is, like Taylor^s Theorem, arranged according to posi-
tive and ascending powers of h, and the other part
R h-^ + Q /i-^ + &c. + B A-(— i) -j- A A-'" (ISO)
is arranged according to negative and descending pow-
ers of h.
2. The coefficient of h~^, in the development of
f.(^x-\-h) by the preceding method, is called the re-
sidual of/, x^ and vanishes for all values of x^ except
those which satisfy (174).
To residuate is to find the residual.
3. Problem. 2 o residuate a given function.
Solution. Let f. denote the given function, and let x^ be
the value of x which satisfies (174). Since R, which is the
residual of ihi.- function by (180), is the coefficient of h™—^ in
(177) the development of ]i"^f.{xQ -\- h) by MacLaurin's The-
orem ; we have by (445 of Vol. I.), if we regard h as the vari-
able,
§ 4.] RESIDUATION. 45
Method of residuating.
provided that after the differentiation we put
hz= 0.
This vanishing of h may be effected in the general form, bj
substituting for h the infinitesimal f, which gives
^ = 1.2.3. ..(m-f)- ^^^^^
4. Examples.
1. To residuate the function (x — a)~^ (x — b)~'^.
Solution, This function becomes infinite of the first order,
when
x z=i a -\- i ;
and infinite of the second order, when
x = b -\- i.
The residual which corresponds to x =. a, is, then,
' i (i)-i (« — 6 + i)-^ — {a — b)-^ ;
and that which corresponds to x :== 6, is
— __(6_a)-2.
1
2. To residuate
(x—a){x—b) (z— c)3
Ans, The residual for xziz a, is {a — b)—^ (a — c)~^,
that for x=b, is (b — a)-'^ (b—c)-^,
that for x=c, IS ^7^-7— r~ -•
46 RESIDUAL CALCULUS. [b. IV. CH. I.
Residuation.
3. To residuate cosec. 2.
Solution. We have
cosec. 2 = GO ,
whenever z=7i;r,
71 being an integer, and the residual of cosec. z is
t 1
t cosec. [n jt -\- 1)^=-
%$
sin. {m^ -\- i) cos.{nn^-j~i'j
1
COS. n n
= ±1.
4. To residuate tang. x.
Ans. ± 1.
5. To residuate Cosec. 2. Ans. 1,
6. To residuate (Cosec. 2)2. Ans. 0.
7. To residuate 2~^ cosec. 2. . ..
-4/15. When 2 = 0, it is ^ ;
% when 2 z= n;rr, it is i {n 7t)~',
8. To residuate x~^ cosec. 2.
Ans. When 2 =: 0, it is 0;
when 2 = n TT, it is i (wti)"'.
9. To residuate — — for any value r^ of z which
2 — z
satisfies the equation
/. z = 00 .
^ 4.] RESIDUATION. 47
Method of residuatins.
%^SoJution. Let f. {x^ -\- i) be infinite of the ?nth order, and
let f . z =r /. z. (z — 3- J™, (183)
80 that f. (Xq -j- i) may be of the zero order, and the required
residual is, by (182),
d:-'.f.{x^+i).(x — x^ — i)- '
1.2.3.... (///.— I)
^ dT.T''r{^o+i) [ (x-x^)-'+{x-T^)-H-{.(x^x ^) -H2+&.c. ]
1.2.3 .... (m — 1)
__ 1 /d:-' f-K + O
1 . 2 . 3 . . . . »i — I \ x
(
3",
+^tSS^' + 'E^?-'H— )(.»•)
But it is evident, from M icLaurin's Theorem, that
rf-\f. (r. + O./"
(185)
J .2.3... (m— I)
is the coefficient of i'"~' in the development of
f.(.r, +{).{" (186)
or, dividing by «", that (IvSo) is the coefficient of {"'-"-'^ in the
development of f . [x^ -\- i). Hence, by this ilieorem,
d^fSJ^^^+i)J^ _ f/— -^ . f . (r^ + _ r/r"-^f.(xj
1.2.3...(m-l)"" 1:2.3.... (m—/i—l) 1.2.3...(77^-/^-l)^ ^
which, substituted in (184), gives, for the required residual,
1.2.3... (m-l)'x—XQ~ 1.2.3... (/«-2)*(x—2q)2~
"T" 1.2 '(x— Jo)"'-2 ' (Z—Xj"'-^ '"(X — Zq)- ^ ^
48 RESIDUAL CALCULUS. [b. IV. CH. I.
Integral residual.
The value of f.x^ is found by the equation (183), which by
(521 of Vol. I.) gives
_ d-{z-x,r _ 1-2 3>- :^^ ,189^
10. To residuate the preceding example, when
z^ -\- ab
f.^ =
{z—a) (z—l>)^'
a^J^ab 1
Ans. When z^^a. the residual is -— . ;
(a — by' X — a
. . b2—2nb—a2 1 b^-^-nb 1
when z==b, it is — — . -\-
(^b—af 'x—b^ b—a ' {x — by'
11. To residuate example 9, when
f .z z=. cosec. z.
Ans. When 2 = w tt, the residual is ri=
X — n 7t
5. The ijiteg?^al residual o( 3. function between cer-
tain limits is the sum of all its residuals contained be-
tween those limits; and the total residual is the sum of
all its residuals.
To residuate from one value of a. variable to another
is to find the integral residual between these values of
the variable ; and to residuate totally is to find the total
residual.
a. The total residual is indicated by the sign ^, and the
^ 5.] RESIDUATION. 49
Notation.
integral residual is denoted by the same sign with letters an-
nexed to it, to show the limits of the residuation ; thus
t(/-^) (190)
is the total residual of /". x ; while
^^■(/•^) (191)
is the integral residual ofjT. x from the limit
X = Xq to X = Xj.
b. The residuation is often limited to those values of the
variable, which render one of the terms or factors of the given
function infinite, as in Example 9 of the preceding section ;
and this is indicated by placing, in double parentheses, the
factor which is thus regarded exclusively of the other factors.
Thus £•((/• ^))-(/'-^) (192)
indicates the residual of (y. x) (/"'. x) with regard to those
values of x, which render f. x infinite. In this way
£-((/-^)) (193)
should be usgd instead of (190) to denote the total residual of
f. X. In the same way
denotes the simple residual of
/.x.(x-r„)^
X — Xq
for the value of x, x z=z x^.
5
50 RESIDUAL CALCULUS. [b. IV. CH. I.
Notation.
6. The variable in (101) may be itself a function of other
variables, as y, z, &c. ; and the residuation may be sought
between the limiting values of y
y — Vo ^"^ y — yxy
and those of z
z = Zq and 2 = 2;^, &c.
and this may be expressed by the form
vy = y,. ^ = ^1, f^-.((/.:r)), (195)
or more simply
it being conventional in what order the limits are placed.
7. Corollary. The preceding notation gives at once, if x' is
a value of x between x^ and x^,
ryl
l7-i(f-^)) = ei •{{/■^)) + 17 ((/■-))• (19-)
8. Scholium, If x' is a root of the equation
/.xzzroo; * (198)
the value of the corresponding residual should be equally di-
vided between the two terms of the second members of (197),
that is, when one of the limits of (191) is a root of (198), one
half of the corresponding residual should be included in the
expression (191).
9. Corollary. If, in (19G), there are only two variables y
and z, and if y is taken to denote the real term of x reduced
<5. 10.] RESIDUATION. 61
Residual of differential.
to the form (1), and z ilie real factor of the imaginary term,
(196) will denote the integral residual for all values of 2",
whose real terms are included between y^ and ?/j, and the
real factors of whose imaginary terms are included between
Zq and z^.
10. Corollary. It is evident from (182) and <5» 5?
that the residual is a linear function ; and found, as it
is, by differentiation, it must by <§> 52 of B. II. be free
relatively to any other linear function^ such as differ-
ence, differential^ &c.
Thus, if the residuation is taken relatively to .r, we
have
L.{{d^-.f.{x,z))) = d^-.C{(f.(x,z))), (199)
52 RESIDUAL CALCULUS. [b. IV. CH. II.
Development of a function, wJiicli has infinite values.
CHAPTER II.
DEV^ELOPMENT OF FUNCTIONS, WHICH HAVE INFINITE
VALUES.
11. Problem, To develop a funclion which has in-
finite values corresponding to finite values of its vari-
able^ in a fortn which may be tised for all values of its
variable.
Solution, het f.x be tlie given function, and let x^ be a
value for which it becomes infinite, so that, if i is an infini-
tesimal, f{xQ -\-i) is infinite of the mih order. Then if we
put f:.xz=fx.{x — x^)'^; (200)
we have f . Xq finite, and (200) can be developed according to
powers of x — x^. We have, by Taylor's Theorem,
d""-^ f X
whence, by (200),
f r // f r d"^-^ f X 1
•^ (x-a:J-^(x-xJ'«-i^ ^1.2.3...(wz-l) x—x^
d'^ f X d"'+^ f X
^1.2.3... 7»^1.2.3.. .. (m+1) ^ °''~ '^ ^
Now the upper line of the second member of (202) consists of
<5> 11.] DEVELOPMENT OF FUNCTIONS. 53
Function, which is always finite, when the variable is so.
terms divided by different powers % — Xq, all of which are
finite, unless
%=x„ (203)
in which case they are infinite ; while the lower line is a func-
tion of X, which is finite in this case. We will denote the
upper line by X^ and the lower line by Y^^ \ and X^ is, by
(188), the residual of
-^-^^ (204)
X — z
when z = x^. (205)
If, then, we denote by Z^ all the other residuals of (204),
when jf.z is infinite; we have, for the total residual of (204),
iS^l^-^X^^Z,. (206)
But by (202) /. z = Xo + Fo J (207)
and therefore /. x — ^. ^^^^-^ = ^o — ^o- (^08)
Now Fq and Z^ are both such functions x)f x that they are
finite when
X = Xq ; (209)
that is, the first member of (208) is a function of a;, which is
finite for every finite value of t, such as (209), for which f»%
is infinite, and if we denote this function by w.x^ we have
f-—l^T^ --.■-• (210)
Hence the second term of (210) is finite for all finite values of
X for which f. x is finite ; and, therefore, -cs x must be Jinite
for every Jinite value of x.
5*
54 RESIDUAL CALCULUS. [b. IV. CH. 11.
Development of a function, which has infinite values.
Hence in the equation
/..= £/i4^ + ..., (211)
the first term of the second member consists, as in (188), of a
combination of terms arranged according to the negative pow-
ers of X — Xq, X — a; J, &,c., while or . x is always finite, and
can usually be developed according to powers of x by Taylor's
Theorem, or by some other simple process.
12. CoroUari/. When the modulus of x is infinite, the first
term of the second member of (211) vanishes, and (211) be-
comes
f.co= cr.x. (212)
13. Corollary. When the first member of (212) is finite for
all values of the argument of z, cr. x is always finite. But it
has been shown, in ^ 81 of B. III., that the equation
-i-rrO, or tjr.x = 00, (213)
uf X
is always possible, unless cr.z is constant, that is, independent
of x; and, therefore, if we put
/. O) = P; (214)
we have
iiif x^F, (215)
and in this case
f.^^lSSl-llj^F. (216)
^ 16.] DEVELOPMENT OF FUNCTIONS. 5$
Development of a rational fraction.
14. Corollary. When f.x is a rational fraction, zj.x is
also a similar rational fraction, because the second term of
(210) consists of the sum of such fractions. But cr. x cannot
have an entire polynoiniai for its denominator, because such a
denominator would vanish for finite values of x, and cj. x would
become infinite. Its denominator must then be constant; that
is, cr. X must be an integral polynomial.
15. Corollary. If, in the preceding corollary, the degree of
the numerator of/", x is greater than that of its denominator,
this function is infinite when its variable is infinite; but if the
degree of the numerator is equal to that of the denominator,
f,x is finite when its variable is infinite ; but if the degree of
the numerator is less than that of the denominator, f, x van-
ishes when its variable is infinite. For if the function is
^x"+/>x"-i 4- &c.
J - — a'x^'+ bx" -'4- 6lc. ^'•^ ' ^
we have /. x = -— — {-- = — (go )"-«' (218)
which is infinite, when n > »',
finite and = -y i= i^, when n z= n', (219)
zero, when n ^ n'.
The polynotTiial uj.x is, therefore, reduced to a constant in
the second case, and to zero in the third case.
16. Corollary. The easiest way of finding zu . x m the case
of § 14, is to reduce the given fraction by division to a mixed
expression, consisting of an integral polynomial, and a fraction
in which the degree of the numerator is less than that of the
56 RESIDUAL CALCULUS. [b. IV. CH. II.
Development of cosecant.
denominator. For this last fraction can, by the preceding
corollary, furnish no part of the polynomial et. ■x, ^vhich must,
therefore, be the same with the polynomial thus obtained by
division.
17. Examples.
1, Develop (sin. a)— ^ by the preceding principles.
Solution. The general expression for the root of the equa-
tion
(sin. x)-i = 00 , (220)
is x = ±/i-^, (221)
in which 7i is any integer at pleasure; and the corresponding
value of the residual of
(sin. z)
X — z
is, by Ex. 9, § 4, if we put
1
i.z =
— 1
d^ . sin. z cos. z
1 1
cos, lire x^nrc
SO that by (21G)
(222)
1 1 1 1,1,1
cosec. X = -, = j r— _ H --
sin.'x X x-\-n X — n x-f-'z/v X — 4 TV
<5> 17.] DEVELOPMENT OF FUNCTIONS. 57
Development of secants.
2. Develop sec. x by the preceding principles.
Anz. sec. 2 = 4 71 1— — -—^ — ., . -7^ + ^^^"^ — v^ — &.c. J
(224)
3. Develop {e + c-^)-i = J Sec. 2;.
Solution, Let a; z= ?/ + ~ V — 1» (225)
and we have, by (89),
e^ z=z cv {co%. z -\- A^ — l.sin. %), (226)
€—''=: e—y{cos. z — V — 1 . sin. z). (227)
Hence the equation e^-|~ ^~" ^= ^^ (228)
involves the two (e^ + e~y) cos. z z= 0, (229)
(ey__e-y) sin. ;:i =z 0. (230)
Hence, cos.^z=0, e'^ — e-y, or e^^—l, y=:0;(231)
z =d^(n + i)n, (232)
and the root of (228) is
^{n + i)n^-l, (233)
7:
If, now, we take
f-=^^e---^=^--^'' (~^'')
we have, by (90 and 92),
1
2x7—1
and the corresponding residual of
X — z
{.r, = ± ;r-7—i ■' i~^^)
(236)
5S RESIDUAL CALCULUS. [b. IV. Oil. II.
Development of a rational fraction.
11 1
^^ ^ o / 1 - -^/. , 1-, —/— 1=^^
'2V— l^=P(^^+=^)-'V— 1 -^/V— i±(2w+l).T
(237)
we have, then,
-L- = ( ' L__\
4. Develop (c"" — e— ^)— ^ =z J Cosec. x.
3-5 4- 1
Solution. Since
(.i- 1)2 (x + 2) z= x3 3z + 2,
(240)
re have, by division,
*'+! -t- 1 3 1 -2^^+9^-5
(241)
(^_,)-2(,+2)-- 1 1 (x_I)2(x + 2)-
Now by Ex. 9 of § 4
^ \\(z-\y^(z^'2)) fx-z 3(x-l)2
I ^^ ^^ . ^24^^^
"T-9(x — 1) 9(z + 2)' ^ "^
<5> 17.] DEVELOPMENT OF FUNCTIONS. 59
Development of rational fractions,
whence, by (216),
^^ + ' -.3+3+ 2
{x — \)2 (x-^2) ' ' 3{x— 1)
2
^ ^^ - ^^ m3^
"^9(x— 1) 9(x + 2)' ^ ^
/*. a;
6. Develop — ^— — ^^ , in which x^., x,,
{x—Xo){x—x^){x — X2)....
&/C. are all unequal, and the values of x, which rentier f. x
infinite, are to be necrlected.
Solution. We have at once
Zlf — r /l^ 1_ (244)
(X-X,){X-X^)... C (((^_3;^)(^_:,J...)) ^_^ V /
(^•O — ^^l)('0 — ^2)--- ^—^
f.x 1
{x^—Xq){x^—X2)(x^—x.J... x — x^
7. Develop
(x+l)(x_2)
^"'* 3C^l)"*"3(x— 2)'
a Develop ^-^---^^.
2 2
(a;_2)2 3(x— 2) ' 3(x+ 1)
BOOK V.
INTEGRAL CALCULUS.
BOOK V.
INTEGRAL CALCULUS.
CHAPTER I.
INTEGRATION.
1. The iyitegral of a given differential is the func-
tion of which It is^th- diifereiitial ; and /Ac integral of a
given finite function is the function of which it is the
differential coefficient.
To integrate is to find the integral. The sign of integra-
tion is /. ; thus
f.d,x = x, f.d.fx — f.x]
f.d,.x = x, f,d,.f.x=fx; (245)
f'di.x = x, f:d:.fx=fx,&.c. (246)
2. Corollary. Since we have
d,i{x + a)=zd,.fx, (247)
for all values of a, it follows that
/.c/,./.xr=/.x + fl, (248)
that is, the integral of a function may have an arhitra-
64 INTEGRAL CALCULUS. [b. V. CH. I.
Increase or decrease of arbitrary constant.
ry constant added to it, and in this form the integral is
said to be complete,
3. Corollary. Any constant may then be added to, or sub-
tracted from the incomplete integral, and the form of the in-
tegral may often be changed by this process.
4. Corollary. If the integral contains a term of the form
log./ ^,
this term may be changed, by the addition of a constant, to the
form
log./. X + log. a = log. {af. x). (249)
5. Corollary. If the integral contains a term of the form
sin.t— ^] X,
this term may be changed, by the addition of a constant, to the
form
sin.C-i] 2 — J^rz: — (Jtt — sin.[-i] x)z=i — cos.[-i] x (250)
or it may be changed into
coseJ-^] -, or into cosj— i]/\/(l-x~) or into - s'mS~^^\/{l-x^),
In the same way, terms of the form
cos.t— ^] 2:, tan.[— ^]x, cot.f— ^] z, sect— ^3 x, &,c.
may be changed into
— sin.[— ^] x, cot.t— 1] X, — tan.-— ^] z, — cosec.t— ^3 2, &:c.
or into
1.1 1 1 '
sect— ^-' -, — tan.'-^^ -, tant— ^1-, cos.t— ^]-, &c.
Z XXX
«J 7.] INTEGRATION. 65
Number of arbitrary constants. Definite integral.
or into (251)
6. Corollary. Since every integration introduces an
arbitrary constant, the nunib(3r of arbitrary constants
in a complete integral must be equal to the number of
integrations.
7. Corollary. The difference between the two values
of an integral, which correspond to two values of its
variable, is called the definite integral from one value
to the other value of the variable.
Thus if Xq and Xj are the limiting values of the variable,
the integral of cl^.f. x from x^ to r^ is, by (248),
(/. ^1 + «) - (/. ^0 + «) =/• ^2 -/. ^0 ; (252)
and it is written
•X.
f
Kd,.f.x=f.x^. (252)
The definite integral is, therefore, independent of the value of
the arbifrary constant ; but the places of the arbitrary con-
stant and the variable are supplied by regarding one of the
limits as arbitrary and the other as variable, thus
J ^(
,d,.f.x=f,x^f.x^, (254)
which gives, by (248),
a — — /. 3-Q, (255)
6*
66 INTEGRAL CALCULUS. [b. V. CH. I.
Integrals are linear functions.
8. Corollary. Since
/^o.rf../.x=/.x„-/.x., (256)
we have, obviously,
fi:=-fi:- (257)
9. Corollary. Equation (246) shows that integration may
be regarded as negative differentiation, that is,
^"^Z (258)
10. Corollary. It is evident, from B. II. '§^^ 51 and
62j that integrals are linear functions^ which are free
relatively to all other linear functions.
Thus we have f.af.x — aj.f x. (259)
11. Corollary. Differ eritials, residuals ^ a7id integrals
are functions which are relatively free.
12. Corollajy. When a function can be separated
into parts connected by the signs + or — , the integral
of the ivhole function is the algebraic sum of the partial
integrals.
This method of integration might naturally be called inte-
gration by parts, but the following is a particular case of it, to
which this designation has been applied technically.
13. If u and v are functions of a variable, we have (Vol. I.
468)
d^.uv z=i udg.v -\- V df.u, (260)
<5) 16.] INTEGRATION. 67
Integration by parts.
whence ud,,.v ^n d^.uv — vd^.u, (201)
and by integration
f,ud^.v z= uv — f.vd,.u; (262)
and when a given differential coefficient can be sepa-
rated into two factors, one of which, d^. v, has a known
integral, the integration can often be effected by the
aid of (262) ; and the application of this formula is
called integration hy parts.
14. Theorem, A definite integral, which is taken be-
tween limits differing by a quantity equal to the differ-
ential of the variable, is equal to the differential of the
integral.
Proof. For the equation (252) becomes, when
3-0 = 2;, a; J r= z + c? X, (263)
by (Vol. I. 421)
fl^'\d,.f.x=:f.{x-\.dx)^f.x^d.f.x. (264)
15. Theorem. \i x^^ x^^ x^^ , , , . x^ are successive
values of x, a definite integral from Xq to ^,„ is equal
to the algebraic sum of the corresponding definite inte-
grals from Xq to x^j from x^ to x^^ 6oc.
Proof. We evidently have
/.^n— /.2:o = (/.2;i — /.Xq) +(/.Z2 — /.2; J
+ (/.^3-/.^2) + &'C. (265)
16. Corollary, Hence if Xq^ x^^ x^^ 6cc. differ by dx,
6S INTEGRAL CALCULUS. [b. V. CH. I.
Change of variable.
the definite integral from x^ to x„ is equal to the al-
gebraic sum of all the corresponding differentials from
Xq to x,i, taken at intervals equal to dx.
17. Scholuim. Propositions 14 and 16 require that the inte-
gral be a continuous function between the limits, and particu-
lar caution must be observed to exclude those cases, in which
the value of the integral varies from positive to negative, or
the reverse, by passing through infinity, so as suddenly to vary
from positive to negative infinity, or the reverse.
18. Theorem. If we have the equation
f.f.x^F.x (266)
and if we substitute for x any function at pleasure, as
' ; (613)
whence
_l^=V(l+tan.= V')=V(l+5-ta,i.= .)
irr\/(l+tan.2 (f — g2 tan.2 (/))=\/(sec.2y — gS tan. 2 ^
=::sec. 9' V (1— e^ sin.2 T) z= sec. 9 ^ . <^, (614)
z= ^ sec. (p cos. a.
In the same way,
1 J tp' cos. &'
(615)
COS. '^j' cos. y' cos. 9'
which, substituted in (611 and 612), give
t =A tan. J (f — 9') ^ g' — =F ^ tan. J (^ — 9)') cos. q (616)
<'=:^ tan. I ( "' + F.(po, &c ; (655)
140 INTEGRAL CALCULUS. [b. V. CH. VL
Elliptic integral of first order.
whence F. i/; . <35 sin. V sin.2 go sin. go 1
^(e.gp) COS." ^ cos. V^ sin.
1 1 1
(666)
^(1 — cos.^r) \/(l-e"sin."^i/;) ^(e.V)
J o^i^'f) J ^ ^(e.g')
V^ -^(e-V') t/ j{c.yj) J j{e.yj)
= F,—F{e.y^); (667)
SO that Fcp and -Fv^ are two functions whose sum is
the function Fi which is called the complete integral^
and the two functions are called complementary with
regard to each other, as well as the angles w andV', upon
which they depend.
142 INTEGRAL CALCULUS. [b. V. OH. VI.
Transformation of elliptic integrals.
99. Corollary. The three angles V^, v^' and ^q^ which
correspond to ;^.^= ^^- ^ — , (682)
'^ cos. (2 X — V) -|- e COS. V^ ' ^ '
i>;f.V_ 2 2{e+\)j".x
^ ip COS. (2 yir — V^)-f-c COS. v^ l-J-e'^-(-2e cos. 2 x
"■ (e + ir(y'.>rr ~(e+l)/'.;^'
(683)
J J^ J
X Dx '^P __
J ■^J
// 2 2 p"
( . M^// -^-Jr. ^"- "^ = 4- ^"- ^- (6B4)
(e+l) ^"./ c4- 1 a/ «
101. Corollary. If ;r, z', -^'q correspond to V^, V', % , we
have the equation
i^."z + i^.";?' — J^.";ro = 0, , (685)
with the condition that
COS. xq = cos. X COS. X' — sin. x sin. / ^" ;ko . (686)
102. Corollary. In the same way in which F' x is ob-
tained from F^^ another function F"x^ might be obtained
from F' Xf and so on, until a series was obtained in which the
values of e, e", e"', &,c. form a series of eccentricities, in which
each differs less and less from unity. It may be shown that
e" differs from unity less than e does, for (677) gives
(689)
1-^ ~(i+v^)'^(i+^)~(i+v^)-^(iW '^'
144 INTEGRAL CALCULUS. [b. V. CH. VI.
Multiplication of elliptic integrals.
in which the factor of 1 — e is evidently less than unity,
and decreases rapidly with the decrease of 1 — e. The value
of F. (p may therefore be made to depend upon a value of
F= {e„. cp„), in which c„ differs from unity by as small a
quantity as we please, and we have
(690)
103. Corollary. In the same way F (ecp), by reversing the
above process, may be made to depend upon the value of
F (cn (fn)) ^^ which e„ is as small as we please. In this case
e" may be found from e by reversing the accents in (677) and
solving the equation with regard to c", which gives
104. Corollary. If we put in (677)
e — tan.2 ^ ?, (692)
we have e" = sin. /5. (693)
105. Corollary. We have by (681, 680, 683),
COS. 2 X = (694)
J -^ — e COS.
yj
(I c2)cos.
■ ^
'J xp — e cos.
.^'
J rl>-\- e COS.
Xfj
c-f-cos.2;r= rj-^^ ^— — ^^=r(^V^+ccos. v)cos.'^ (695)
J" x^
1+6
<§> 106.] CIRCULAR FUNCTIONS. 145
Reduction of elliptic integrals.
/V^ ( J yp-\-e COS. v^)2_ /» ^' 2(^T/;) ^4-^ e cos. ^p J yj-( l-e^)
/^ /I -^ /•V 1 — e Z*'/^ « COS. 1//
T+c ""^ 2"^ "^^ 14-e
= ^-ni-e)F^+'-f^, (696)
which may serve to deduce the value o^ E {e . (f) from those
of E.(e„cp„), in which c„ is very small, or differs but little
from unity.
106. Corollary. Potential functions may be applied to the
hyperbola very nearly in the same way in which circular func-
tions have been applied to the ellipse. Thus if we put
A Cos. (p = X, B Sin. (p = y, (697)
X and y are the coordinates of the hyperbola, of which the
equation is
(i)-(fy->- («^«)
The length of the hyperbolic arc is, by putting
^ = v(l+2), (699)
s=Bfs/ (l + c^ Sin.s^)).
If we let
^ (c (t) = \/ (1 + e^ Sin.'-J q) (700)
ar {ecf)= /'^^(eT), (701)
»/
we have s ^= B ;j {c(f). ("62)
107. Corollary. The condition that the point of contact in
(fig. 2) is upon another hyperbola which has the same foci with
13
146 INTEGRAL CALCULUS. [b. V. CH. VI.
Hyperbolic integrals.
the given hyperbola, is expressed algebraically by the equa-
tion
Cos. )q := Cos. (f> Cos. cp' — Sin. cp Sin. (f'^ r ^o t (703)
and corresponds to the equation
3" (f'+ 3" ' — a- s usually depends upon many variables,
t, X, 1/, &LC., and their differential coefficients, in such a way
that, if t is taken for the independent variable, any change in
the functions by which a-, y, &,c. depend upon t, gives the
equations
ds,=Ds,.dt,, Ss^z=:D So . ^t^, (743)
d Dsz=X3x+YSi/+&Lc. -\-X'S Dx-\~Y'3Dij-{-&.c.
+X" i 2>2 a;_|_ Y" $ Z>2 3,-|-&,c. ; (744)
154 INTEGRAL CALCULUS. [b. V. CH. VI,
Method of Variations.
which, substituted in (742), give
-\- X" 8 D"^ xJ^&LC. +F,Jy+&c.)=0. (745)
But, by (262),
/. X' d D x=f. X' D 5 x=X' d x—fDX' . $x (746)
/. X" d D^xz=f. X" Z/2 8xz=iX" D 8 X—fDX". D $ x
—X' D d x-D X". d x+f i>2 X". S X, &LC. (747)
The terms in the last members of (746 and 747), which are
not under the sign of integration, must, in passing to the defi-
nite integrals, be referred to the limits of integration But it
must be observed, that the variations in (746 and 747) are
taken upon the supposition that the independent variable t
does not itself vary, and that only the functions vary, by which
Xj 7/, fcc. are connected with it ; whereas the limits of inte*
gration may themselves necessarily vary with a change of this
function, and therefore t^ and t^ are supposed to vary. If,
then, 8' Xq , ^'^Q, iS^c. denote the variations arising from
the change of the functions, the values of the complete varia-
tions are
3x=zd'x^-\- Dx^.St^, 6lc. (748)
whence 8' x^z=z ^ x^ — D x^ J t^, &c. (749)
Hence (746 and 747) give
f^^^ XdD x—X[ 3'x^—X'o -5' ^o—fl^ DX.dx (750)
^^X'5D^xz=X[ Dd'x^—DX'; d'x^—X'^DS'x^
f
— DX; 6'x^+J^^^^ D^X". <5 X, &,c. (751)
in which S' is given by (749).
§ 116.] RECTIFICATION OF CURVES. 155
Method of Variations.
These equations, substituted in (745), give
+(X;- &c.) D S'. x^—iX'^-SLc) D ^' z^-f &,c.
(752)
The terms of (752), which are under the sign of integration,
express a variation which belongs to each point of the curve
independently of all the other points, and which must, there-
fore, be equal to zero for each point ; which gives the general
equation
{X—D X'-\-D^X"—&LQ,) S X+&C. = 0. (753)
The variables, t, x, y, &c. may be bound together by some
conditions, represented by the equations
i = 0, M=:0, (754)
in which L, M, may be functions of t, x, y, &,c. The varia-
tions of these equations will then give linear equations between
^Xj ^y, &:,c. from which the values of some of the variations
iXj <5y, &c. can be determined in terms of the others. These
values, substituted in (753), will reduce the number of varia-
tions in (753) to the smallest possible number, and those which
remain will be wholly independent of each other, and there-
fore their coefficients must vanish. The equations, thus ob-
tained from making these coefficients equal to zero, will be
the required equations of the shortest time.
If, in addition to the equations (754), the limits of the curve
are subject to peculiar conditions ; these conditions, with those
of (754), referred to the limits of the curve, may be combined
with the terms of (752), which are not under the sign of in-
tegration, and the equations for determining the extreme points
/
156 INTEGRAL CALCULUS. [b. V. CH. VI.
Maximum or minimum of definite integrals.
of the curve may be found by the same method by which the
equations of the curve itself are found.
117. Corollary. The preceding process for finding
the minimum of (741), may be apphed to finding the
maximum or minimum of any definite integral, such
as
'■• V, (756)
by changing in the various formulae D s into F.
118. Corollary. The number of the variations ^x, Sy, &lc.
determined by (754), is plainly equal to the number of the
equations of (754). The number of the variations left unde-
termined, therefore, in (753), and consequently the number
of equations obtained from (753), is equal to the number of
the variations not determined by (754). The whole number
of equations then of the required curve, is equal to the whole
number of the variables 3", y, z, & ^r^r^s
^'"^ IT ^UT ('^')
and
1)2 o __ ^ n L _ the area ah c d .^^^.
'•" ~ dl ^ dl.dm ' ^ '
But, if a z=i the angle h a c, ^
5' = an arc of Z Z', > (801)
s" = an arc of 3Im ; j
we have the arc a b z= d s', the arc a c =: ds", (802)
the area ab c dz=i sin. a d s' . d s'
and, since m is the only variable in s', and / the only variable
m s'
D],^o — sin. aD^s' . D^ s", (803)
«5> 134.] QUADRATURE OF SURFACES. 167
Area of a curved surface.
and the accents may be omitted in (802) without any ambi-
guity. Hence
c =ifj,, sin. aD^.s, Di.s; (804)
in which D^s and DiS may be taken directly from the gen-
eral expression for D s, and a is the inclination of two lines
drawn through a point, in such a way, that for the one / is
constant, and for the other m is constant.
132. Corollary. ^ When the surface is plane, (570) of vol. 1
gives for rectangular coordinates,
l>,s=l, />, s=l, (805)
and it is obvious that a is a ri^ht angle ; whence
«=/,/,.!=/,. a: =/..y, (806)
or supplying the place of arbitrary constants by the form of
definite integrals,
" =fi fi: ' =/:: ^^^-y^) =f'i: (^.-^»)' (^«')
in which the values of a-^ Xy 7/0^1, ^^^ determined by the
bounding curve.
133. Corollary. When the surface is plane, (574) of vol. 1
gives
DcfS=r, DrS=\, (808)
and « is a right angle ; whence
o =.f^f^ ^r^f^,r^P=^ if^p . r^• (809)
or
134. Corollary. When the surface is curved, let Y denote
the inclination of the tangent plane to the plane of x y, and,
168 INTEGRAL CALCULUS. [b. V. CH. VII.
Area of a curved surface.
since the projection of a surface is equal to the product of the
surface by the cosine of its inclination to its projection, (806)
gives
o=fJ^.sec.Y. (811)
Hence, by (600) of vol. 1, where
V—0 (812)
is the equation of the surface, *
= fjy ■ V (D. z^ + ^,==' + !)• (813)
135. Corollary. When the surface is developable, it may
be supposed to be developed into a plane, and its area found as
that of a plane surface; or it must give the same result to
refer the surface to axes, drawn upon it in such a way, that
they would be straight lines when the surface was developed,
and the rectangular coordinates would then be the length of
the shortest lines, which would be drawn upon the surface to
two of these axes, which would be perpendicular to each
other.
136. Corollary. When the surface is one of revolution, the
notation of § 129 gives, by § 134,
a^f,^f^.u^{D^z'^+\); (814)
and if 5 denotes the arc of the generating curve,
"" =hfu 'U DuS =fcpf, ,uD,.s =fcff, . u. (815)
137. Corollary. When the surface of revolution is included
between four curves, of which two are the intersections with
the surface of two planes which are perpendicular to the axis
Ǥ> 140.] QUADRATURE OF SURFACES. 169
Quadrature of a surface of revolution.
of revolution, and the other two are the intersections with the
surface of the planes, which may be called meridian planes,
because they include the axis of revolution, and which are in-
clined to each other by an angle Vg j (^1^) gives
138. Corollary. If another surface of revolution were gen-
erated by the revolution of the arc in the preceding section,
about an axis at the distance h from the former axis, and
farther from the arc, so that for this new axis we have
u' = u + h, (817)
(816) gives the value of the corresponding surface
o' = ^^fl^^ {u D^^ s + b D,, s)
— a^bcp^ (Si—So). (818)
139. Corollary. Had the second axis been upon the oppo-
site side of the arc, we should have had
u" z= b — u (819)
o" = b^,{s,-s,)-o. (820)
140. Corollary. A curve AB A' B' (fig. 8) is said
to have a centre when there is such a point that any
chord, such as A A\ B B', &c. which passes through
it, is bisected by it ; and such a chord is called a diam-
15
170 INTEGRAL CALCULUS. [b. V. CH. VII.
Surface of a ring.
eter. The surface generated by the revolution of such a
curve about an axis C C' which does not intersect the
curve, is called mi angular surface^ or, simply, a ring.
The notation
S =z the perimeter of the generating curve
ABD AD' A,
o =: the surface which would be generated by
the revolution o( D B A D' about the di- ^ (821)
ameter D D' parallel to -CC",
h z=. the distance of the axis C C' from the cen-
tre,
gives by (818 and 820) for the whole surface of the ring,
=:2b^ S. (822)
141. Problem. To transform the differential coefficient of
a surface from one system of variables to another.
Solution. Let / and m be the given variables, and let the
second member of (803) be denoted by //, that is,
Dl^.o^zH (823)
If, then, only one of the variables m is to be changed, and
t is to be introduced instead of it by means of the equation
M = m, (824)
in which M is a given function of I and t ; we have
D].,.o — D,D,.o^D^D,.o.D,.m
— Dl^.G . D,3I=z H D,.M. (825)
«J 144.] Q,UADRATURE OF SURFACES. 171
Transformation of differential coefficient.
If the Other variable / is also to be changed, and u to be in-
troduced instead of it, by means of the equation
/ = L, (826)
in which Z« is a given function of t and u ; we have
D] ,o^HD,.M.D^L, (827)
142. Corollary. If J/, in equation (824), instead of be-
ing a given function of t and /, were a given function of t and
u, u might be eliminated by means of (826). It is more con-
venient, however, to eliminate its differential coefficient only
from Di . 31, after having determined this differential coef-
ficient by means of (826). Thus the differential of (826)
relative to t is, by regarding w as a function of t,
= D,L + D,L . D,u, (828)
whence ^^ , ^ _ g^ , (829)
and (824) gives
= —j^^ . (830)
But 771 is obviously to be substituted for 31 in (827), whence
we have by (827 and 830),
Dl^ azzzH (D,3I.D^,L—D,,3I.D,L). (831)
143. Corollary, The two preceding articles may be
applied to the transformation of any second differential
coefficient of two successive variables.
144. Examples.
1. To find the area of the segment of an ellipse included
between two parallel lines.
172 INTEGRAL CALCULUS. [b. V. CH. VII.
Transformation of differential coefficient.
Solutio7i. Let the ellipse be referred to conjugate axes, as
in (74) of vol. 1, in which the axis of ?/ is drawn parallel to
the given lines ; and (804 and 807) give, since in this case
y„ = -y, (832)
is the ordinate y of the ellipse, if « is the angle of the axes
= 2 f""^ . 7/ sin. «. (833)
If, now, we take 5 so that
X ^^ A cos. 5, (834)
we have y z=. B sin. ^, (835)
D^,x = — A sin. ^, (836)
a=2sin. « r ^ A B sm.^ &z=zA J5 sin. « /" (i — cos. 2 &)
J &i */ ^1
=^jB sin. a [6^—6^—^{s\n. 2 ^q— sin. 2 ^ J] (837)
=■ J sin. « (corresponding areaof asegmentof a circle whose
radius is yl).
2. To find the area of a sector of an ellipse, when the ver-
tex of the sector is at the centre of the ellipse.
Solution. In this case (834 and 835) give, when A and B
are the semiaxes,
r COS. (p =z A cos. Q,
B
tan. (p zzz — tan. &,
A
t^Dq. pzzzA B ;
whence, by (810), putting zero for r^ ,
c = iAB(\-i,). (841)
r sin. (f z= B sin. 6,
(838)
^ B C0S.2 (p
T) (p —
^ A C0S.2 a '
(839)
(840)
<5> 144.] QUADRATURE OF SURFACES. 173
Area of elliptic segment and sector.
Corollary. The whole area of the ellipse \s, n A B. (842)
3. To find the area of a sector of an ellipse, when the ver-
tex of the sector is at a focus.
Solution. If the origin of coordinates is at the focus, (834
and 835) give
r cos. (f=iX7^A cos. ^ — A e=A (cos. & — e) (843)
r sin. (fz:zi/^zzB sin. & (844)
B sin. & ,r.4^.
D6,^?. (lz:i££!_!L-^^ (S46)
A (cos. <3 — e)2 ^ ^ '
r^ D6cp = AB . (1— e cos. &), (847)
whence, by (810),
<^=zA B . [6^—6^—e (sin. ^^— sin. dj]. (848)
4. To find the area of the hyperbolic segment included be-
tween two parallel lines.
A?is. If the hyperbola is referred to conjugate axes as in
(90) of vol. ], in which the axis of ?/ is parallel to the given
lines, if y is the angle of the axes, and if ^ is taken so that
xz= A Cos. ^ y = B Sin. ^, (849)
the area is
a = J ^JBsin.y (Sin. 2^j— Sin. 2 ^0+2 ^1—2 5 J. (850)
5. To find the area of the hyperbolic sector, the vertex of
which is at the centre of the hyperbola.
15*
174 INTEGRAL CALCULUS. [b. V. CH. VII.
Area of hyperbolic segment and sector.
Ans. With the notation of the preceding example, the area
"^^B ("-«„)■ (851)
in which A and B are the semi-axes,
6. To find the area of the Ijiyperbolic sector, the vertex of
which is at one of the' foci.
Ans. With the notation of the preceding example, the area
is
oz=AB [(^— ^o) — ^ (Si»- ^— Sin. d J]. (852)
7. To find the hyperbolic segment included between an
asymptote, the curve, and two straight lines drawn parallel to
the other asymptote.
Solution. It is convenient, in this case, to take the two
asymptotes for the oblique axes, for which « and ^ in (86) of
vol. I. must have the values
tan. a z= - , tan. ^ = — - ; (853)
whence (86) gives for the equation of the hyperbola, referred
to its asymptotes,
xy:=:zl{A^+B^). (854)
The area of the required segment is, then, by (807, 853 and
854), if the axis of y is the asymptote parallel to the given
lines,
. ^ px. 2AB px.
^ «/ 3 ^
JAi3 1og. ^. (855)
<5> 144.] Q,UARDATURE OF SURFACES. 175
Area of parabolic and cycloidal segments.
8. To find the area of the parabolic segment included be-
tween two parallel lines.
Ans. If the parabola is referred to oblique axes as in (100)
of vol. I, of which the axis of 3/ is parallel to the given lines,
and if « is the angle of the two axes, the area is
""^l (i/i 2:1— yo 2:0). (856)
9. To find the area of the parabolic sector, of which the
vertex is at the focus.
Ans. If P is the distance from the vertex to the focus, if
the origin is at the focus and the angle y counted from the
vertex, the area is
— 2P (tan. J ^1— tan. J 'o). (857)
10. To find the area of the segment included between the
curve, the axis of x, and two lines drawn parallel to the axis
of y, of the curve known as the parabola of the order a, which
has for its equation
2/ = Ax\ (858)
Ans. o=^J^rZhlo_, (859)
11. To find the are of the segment of a cycloid given by
equations (130 and 131) of vol, 1, included between the curve,
the axis of x, and two lines drawn parallel to the axis of y.
Ans. ^=722[|.(^,-5o)-2(sin.^-sin.^o)+l-(sin.2^i-sin.2 6o)]. (860)
Corollary. The whole area included between a branch of
the cycloid and the axis of z, is
C = ^7VR^^ (861)
= three times the area of the generating circle.
176 INTEGRAL CALCULUS. [b. V. CH. VII.
Area of sectors of spirals.
12. To find the area of the segment of a cycloid, which is
included between the curve and a line drawn parallel to the
axis of y.
Ans. o—R^[(^n—&) (1+2 cos. (3)_2sin. a— Jsin. 2^].
13. To find the area of a sector of the spiral of equation
(133) of vol. 1, when the vertex of the sector is at the origin.
14. To find the area of a sector of the hyperbolic spiral, the
equation of which is (135) of vol. I, when the vertex of the
sector is at the origin.
Ans. o — 2n'^R^ /_i___iy (864)
15. To find the area of a sector of the logarithmic spiral, of
which the equation is
r = a e , (865)
when the vertex of the sector is at the origin.
Ans. o^^a'^ie ^' — e °). (866)
16. Given the area <^ of a surface included between any
lines whatever, the combination of which considered as one
line which in general is discontinuous, is represented by the
equation
F.{T,y)=0, (867)
to find the area a of the surface bounded by the line or system
of lines
Ans. a'— aha. (869)
<§> 144.] QUADRATURE OF SURFACES. 177
Area of a zone of an ellipsoid.
Corollary. If a and h are equal, the surfaces are similar,
and (869) gives
o'=a^a- (870)
that is, the areas of shnilar surfaces areproportional to
the squares of their dimensions.
17. To find the area of the zone of an oblate ellipsoid of
revolution which is included between two planes drawn per-
pendicular to the axis of revolution.
Solution. Let the notation be that of Example 1, of § 130,
and (816) gives, for the area,
7 ^^^^^
Let the angle w be so taken that
B Sin. ^== Ac COS. 6 ; (872)
and we shall have
— Ae sin. 6 Doj.^^B Cos. oj; (873)
whence
2 -ft jB2
/:
'° . COS.2
e ., 1
= - / ° . (1+CoS. 2 a,)
= ^^ [(%--i) +4 (Sin. 2c.^-Sin.2 .J]. (874)
18. To find the area of the zone of a prolate ellipsoid of
revolution which is included between two planes drawn per-
pendicular to the axis of revolution.
178 INTEGRAL CALCULUS. [b. V. CH. VII.
Area of a zone of a hyperboloid.
A/is. With the notation of Example I, of § 130, and put-
ting
COS. oj ^= e cos. &, (875)
the area is
" = "^^ [(- --'o) + i (sin- V -s'»- 2 %)]• (876)
19. To find the area of the zone of the hyperboloid of revo-
lution formed by the revolution of an arc of an hyperbola
about the transverse axis,
Ans. If the equations of the generating hyperbola are
xz=z A Cos. ^ !/ = B Sin. &, (877)
and if w is taken so that
e Cos. & z=z sec. «», (878)
the area is
_nABrsm.'^^ sin.o'o , ,_ tang. (45° +^ co )-|
^-^- L^^r^^'o+^'t^^ii:(45^+i^)J-^
20. To find the area of the zone of the paraboloid of revo-
lution, included between two planes, which are perpendicular
to the axis of revolution.
Ans. If P is the distance from the vertex to the focus, and
if & is so taken that
y = 2P tan. 5, (880)
the area is
a = f T P2 (sec.s^i— sec.^dj. (881)
21. To find the area of the zone generated by the revolution
of an arc of a parabola about the axis of ?/ of the preceding
example.
Ǥ> 145.] QUADRATURE OF SURFACES. 179
Area of a zone generated by the arc of a cycloid.
Ans. If ^ is taken so that
X -{- P = P sec. 6, (882)
and if o' is the value of o in (879),
the area is P2 e
A B
(883)
22. To find the area of the zone generated by the revolu-
tion of an arc of a cycloid about the axis of x in (130) of
vol. 1. The arc is supposed to commence with ^.
A?is. With the notation of equations (130 and [31) of
vol. 1, the area is
G=l6 7tR2 {2_2 COS. i 5 — ^ sin.2 J 6 . cos. J 5). (884)
23. To find the area of the zone generated by the revolu-
tion of an arc of a cycloid about the axis of y in (131) of
vol. 1. The arc is supposed to commence with 6.
Ans. With the notation of the preceding example, the area
is
a = 16 nR2 (sin. J ^— J & COS. J ^ — ^ sin:3 J 6). (885)
145. Problem. To find the area of the zone gen-
erated by the revolution of a given arc of a plajie curve
about an axis in the same plane with the arc^ when the
areas of the two zones are known which are generated
by the revolution of the arc about two axes in the plane,
which are perpendicular to each other.
Solution. Let the two perpendicular axes be those of x
and y, and let the given areas be, by (816),
180 INTEGRAL CALCULUS. [b. V. CH. VIL
Greatest or least surface.
o'=r 2 TT
o"—2
r^'.x. (887)
Let the new axis be inclined to the axis of a; by an angle «, and
pass at a distance a from the origin, and the required area is
0=1:^2^ f ^ ' {y eos. « — X sin. « — a)
— _j_ 2 TT [a' cos. « — o" sin. a — a {s^—s^)], (888)
in which that sign is to be adopted which renders the second
member positive.
146. Problem. To draw the curve line subject to
given cojiditionSj which includes a maximum or mini-
mum surface.
Solution. This problem, like that of § 116, involves the
maximum or minimum of a definite integral, and is therefore
solved in a similar way, by the method of variations. There
is, in this case, however, a double integral, and the first inte-
gral refers evidently not to disconnected points, but to the
bounding lines of the surface, so that the determination of
these lines may involve the method of variations, even when
the general form of the surface is given. The determination
of the form of the surface will admit of more lucid dis-
cussion in a chapter upon the curvature of surfaces, and
the present chapter will be confined to the consideration of the
bounding line.
The equation of the surface being given, the form of its
second differential coefficient is known, and is independent of
<§> 148.] QUADRATURE OF SURFACES. 181
Greatest or least surface.
the limiting lines, so that an integration can be directly per-
formed, and the required integral be reduced to the form (756),
and the process of finding the maximum or minimum becomes
identical with that of § 1 IG.
147. Corollary. A kind of equntion of condition is often
connected with this problem, wholly different from those refer-
red to in § 116. Each of the equations (754) is an equation
which is satisfied by the coordinates of each point of the re-
quired curve, and is thus equivalent to an infinite number of
equations. But an equation, of the class here alluded to, is a
single equation, involving the coordinates of every point of the
curve. An instance of such an equation is the one which
expresses that the bounding curve must be of a given length,
or that the definite integral (741) must have a given value.
All equations of this kind would appear to depend, neces-
sarily, upon definite integrals, and they may be introduced into
the equation of maximum or minimum for the purpose of elim-
ination by the method of § 119. It must be observed, how-
ever, that the multipliers ;., //, &c , of these equations arc
always constant. For each of these equations does not deter-
mine any relation between $r, $y, &c. which is applicable to
each point of the curve, but only a particular relation by which
one of the variations, as ^x, may be determined for one of the
points in terms of the values of the variations for all the points.
The corresponding multiplier '', therefore, must have that par-
ticular value which shall cause this single value of ^x to disap-
pear from the equation ; that is, ;. must be constant.
148. Examples.
1. To find the plane curve which, having a given length,
encloses the maximum area.
16
182 INTEGRAL CALCULUS. [b. V. CH. VII.
Greatest or least surface.
Solution. The function to be a maximum is, by (806),
'^ y, (889)
and the function (566) is to be constant. Hence if A is the
constant multiplier introduced for the purpose of elimination,
the equation is, by the reduction of § 121,
1 - ^ D. (g^ ) = 0, (890)
or by the notation of § 148 of B. II., and by (577 and 609
of vol. 1,
= 1+^1), .COS. „ (891)
= DrX -\- A Dr COS. v
— sin. r Dys — A sin. ^ (892)
Az:z DrS — Q; (893)
that is, the curvature is constant, which is the property of no
Other curve than the circle ; the required curve is, therefore, a
circle ; which has, already, been proved in the Elements of
Geometry.
2. To find the plane curve which, being drawn from one
given point to another given point, and having a given length,
encloses the maximum area between the curve itself, its two
extreme radii of curvature and its evolute.
Solution. By adopting the notation of the preceding article,
the required area may be expressed in the form
1
. e~ ; - (894)
that of the arc will be
s
=y ;:■ . c. (895)
§ 148.] QUADRATURE OF SURFACES. 183
Greatest or least surface.
Equations (576, 577 and 009) of vol. 1, give
Di X z= sin. r Di 5 = o sin. ,, (S9G)
Dy 7/ Z=Z COS. V Di S = n COS. i . (897)
The given differences of the coordinates of the extreme points
of the curve are, then,
X, — 2, = fl' . Q sin. r, (89S)
ft/ '
!/i—!/o——t ' ' • ^ COS. ,. (899)
If, therefore, A, B, C are the constant multipliers of (895,
898 and 899), introduced for the purpose of elimination, the
equation of the maximum or minimum is
o o _|- ^ -|_ J5 sin. r — C COS. r = 0. (900)
Let H and « be taken so that
B z=zH COS. cc^ C=Hsm.ai (901)
and (900) becomes
2q + A +Hsm. (»' — a) = 0; (902)
and by putting
r' = v — «, (903)
2Q + A + Hsin.r' = 0; (904)
which shows that (900) may be reduced to the form (904),
from which the term containing cos. ^ disappears, by merely
changing the direction of the axis of x. It does not, then,
diminish the generality of the solution to put
0=0; (905)
by which (900) becomes
2 § + ^ + J5 sin. V = 0. (906)
184 INTEGRAL CALCULUS. [b. V. CH. VII.
Greatest or least surface.
Tiie curve is easily expressed in rectangular coordinates by
the equations
2 = ^ A COS. r -\-l B sin. 2 ' + ^ r, (937)
ij — ^A sin. '■ — ^ B COS. 2 »•. (908)
Corolla nj. When the extreme points are not fixed, tlie
equation (900) becomes
'2 c + ^ = ; (909)
that is, the curve is a circle.
Corollary. When the length of the curve is not given, the
equation (906) becomes
"Zq + B sin. 1=0; (910)
which is, evidently, from example 3 of § 151 of B. II., a
cycloid.
<5> 149.] CURVATURE OF SURFACES. 185
Curvature of a surface in any direction.
CHAPTER VIII.
THE CURVATURE OF SURFACES.
149. Problem. To find the curvature of a given
surface at any point in any direction.
Solution. Let the tangent plane to the surface at any one
of its points be taken for the plane of the coordinates x and y,
so that the normal may be the axis of z. We have, then, at
this point,
D,z=0, D,^z = 0; -(911)
and if q^ and c^ are the radii of curvature at the point of the
intersections of the planes ofocz and 1/ z with the surface,
equation (610) of vol. 1 gives
-^^Dlz, ^-=I>lz. (912)
The radius of curvature (? of a section made in any intermedi-
ate direction by a normal plane, which is inclined to the axis of x
by the angle «, is derived from the equation
— = Bl z, (913)
if u denotes the distance of a point of the curve of intersec-
tion from the axis of z. But the coordinates of one of these
points are
xz=iu cos. «, y =^u sin. « ; (914)
16*
186 INTEGRAL CALCULUS. [b. V. CH. VIII.
Directions of greatest and least curvature.
whence, in general,
D^ z = COS. » D^z -j-sin. ^ DyZ, (915)
-=zDlz r= COS.- a Biz 4-2 sin. « cos. « B'l_yZ -\- sin.^a Dl z
C0S.2 a , sin.~ « , ^ . ^o
= 1 1- 2 sin. "COS. oc Dl,y z. (916)
150. Corollary. The radius of curvature (/, in a direction
perpendicular to that of ^, is given by the equation
1 sin."« , cos.2« ^ . _.
—=: 2 sin. « COS. a D'i y z. (91 / )
151. Corollary. The sum of (916 and 917) is
-+T = -+-; (91S)
that is, the sums of the reciprocals of the two radii of
curvature of any two perpendicular sections at a given
point of a surface is a constant quantity.
152. Corollary. If Q were the maximum radius of curva-
ture at the point, o' would obviously be the minimum radius of
curvature ; whence
The directions of greatest and least curvature of a
surface at any point are perpendicular to each other.
153. Corollary. The difference between (916 and 917) is
— — i=:cos.2« f - — -^ — 2sin.2«Z>|.. ^, (919)
Q' Q \Qy Qz /
and in the hypothesis of the preceding corollary, the first mem-
<§> 155.] CURVATURE OF SURFACES. 1S7
Motion of point of contact in direction of greatest or least curvature.
ber of (919) is a maximum. The differential coefficient of
the second member, taken with reference to «, must be equal
to zero, that is, ,
= sin. 2 '^(- — ,7 ) + 2 cos. 2 a D\.yZ. (920)
The sum of (919) multiplied by cos. 2 «, and of (920) Uy
sin. 2 a, is
cos.2«(^--^=:---. (921)
\ '' 'I / " Or
Hence, from (918),
cos.2 « sin.'2 «
from which the curvature of the surface can be found
in a direction inclined by the angle « to the direction of
maximum curvature.
154. Corollary. One half of the difference between (919)
multiplied by sin. 2 «, and (920) by cos. 2 «, is
D-j,.y z z= — 1 sin. 2 «
Q-D- '-'
155. Corollary. For the direction of the maximum or min-
imum, « is zero or a right ajigle, and, therefore, for either of
these directions,
D%.yz = 0- (924)
that is, with a small motion of the point of contact in
the direction of the greatest or least curvature, the tan-
gent plane rotates about a line perpendicular to the
direction of the motion of the point.
188 INTEGRAL CALCULUS. [b. V. CH. VIII.
Direction of no curvature.
156. Corollary. When « is half a right angle, (921 and
922) give
Qy = Qx, (925)
l=l=Wi+M. (926)
157. Corollary. When the values of Q and (?' have opposite
signs, neither of the corresponding curvatures is strictly a min-
imum, but the two curvatures are the greatest curvatures in
opposite directions. There are, in this case, two intermediate
directions of no curvature, corresponding by (922) to the
values of «,
tang. «r=±\/(— ^yV (927)
The sections of the surface, made in these directionSj
have a contact of the second order with the tangent
plane, and correspond, in general, to points of contrary
flexure.
158. Corollary. In the case of a point of contact for which
the greatest and least curvatures are in opposite directions and
equal, we have
Q^-Q'; (928)
whence, by (918),
Cz = — (?,; (929)
that is, the curvatures iu any two directions, which are
perpendicular to each other, are equal and opposite.
We have also in this case, by (927),
« — zb 45° (930)
for the angles, which the directions of no curvature make with
the direction of greatest curvature.
<§> 160.] CURVATURE OF SURFACES. 1S9
Curvature of a section which is not normal to the surface.
159. Corollary. If the curvature were required of a sec-
tion, the plane of which did not include the normal, it might
be found by referring the surface to an oblique system of co-
ordinates, of which the tangent plane was the plane of xy,
the cutting plane that of xz\ the axis of x being the intersec-
tion of these two planes, and the axes of y and z' being per-
pendicular to that of X. This system might be obtained from
the rectangular one, which has the same axes of 2: and y, but
in whicli the axis of z is the normal, by putting
& = the inclination of the axis of z to that of z' , (931 )
= the complement of the inclination of the given plane to
the tangent plane,
which gives 2; =r 2' cos. ^, (9i.5'2)
D: z = Dlz' cos. <5 ; (9:3:i)
or, by putting Q^ =z the radius of curvature of the inclined
section
-=i cos. 5, (934)
(1^ = COS. &. (935)
IGO. Corollary. If the axes, in the preceding corollary,
were rectangular, that of y being perpendicular to the given
plane, and those of x and z situated in any way whatever in
that plane, equation (610) of vol. 1 gives
L — - — -' - — ^^ . . (936)
If we put
the anufle of C'^ and z
y =z the angle of o and z, ^ v ' )
and observe that the plane of i' and 'jj is perpendicular to that
190 INTEGRAL CALCULUS. [b. V. CH. VIH.
Curvature of any point of the surface.
of ;:; and z, so that if a sphere were described with the point
of contact for the centre, the arcs &, t, y would form upon the
surface a right triangle, of which y was the hypothenuse, we
have
cos. y =: cot. r COS. &. (938)
But the comparison of (81 1 and 813) gives
sec.y = ^[\+(D,zy- + {D,zf], . (939)
and we have, obviously,
sec. Tz=V[l+(/>x2)2]; (940)
whence
1 1 sec. r Dl . z
Q Q^ sec. Y l+(Z>x zf
Dl.z 1
. cos. y
1 + {D. zf • V [1 + {Dy ^f+ {D. ^f]
■ (941)
161. Corollary. The curvature of a section of the surface
made by a plane which includes the axis of z, and is inclined
to the plane of 2; x by the angle £, may be found by the formula
in which w = the distance of any point of the section from
the axis of z,
whence
a;=M cos. c, y ^=-u sin. s ; (943)
i>„a;=:cos. «, X>„yz=sin. «; (944)
D^ 2=cos. t .D^z-\- sin. « Dy z, (945)
Dl z=cos.2 a . Dlz+'^ sin. « cos. £ DJ .,2+sin.2 a . I>2 ^ j (946)
1_ _ {Dlz + 2 tan, e . Dl,yZ-\-i^n.^ s . Dl z) cos, y
Q "~l+Z>^22_|.2tan.eZ>,2;i>y2;+(l+i>,%2)tan.2e' ^ ^
<§> 163.] CURVATURE OF SURFACES. 191
Curvature of any part of the surface.
and since the coordinates x, y, z do not themselves occur in
this value of the reciprocal of the radius of curvature, but
only their differentials, (947) is applicable to any point of the
surface, and to any direction of the curvature, it being ob-
served that s is the angle, which the plane, dravi^n through the
axis of z and parallel to this direction, makes with the plane
of xy.
162. Corollary. When the plane which is parallel to the
required direction of curvature is also parallel to the radius of
curvature, (601, 598 and 599) of vol. 1 give
cos. fi Dy Z /rv40x
tan. £ = ■ = — ^ ; (948)
COS. « IJ^ z
whence the product of the denominator of (947), by Dx z^,
becomes
D^z^ ^ Dxz^ + 2D. %2 DyZ^ + DyZ^ + D^z^
= {D, z^ + Dyz') (1 +Dzz2 + Dy z2)
= {D^z^ -{- Dyz2)sec.2Y; (949)
and (947) becomes
1 Drz^ Dlz+2 D, z DyzDl^y^z-^Dyz'Dlz
Dx z^+Dy z'
cos.3y.(950)
163. Corollary. When the direction of curvature is per-
pendicular to that of the preceding article, the plane which is
parallel to it is also perpendicular to that of the preceding
article ; whence, in this case,
Dxz
tan. i' -Z cot. « = =::^ — ,
U z
and (917) becomes
19-2 INTEGRAL CALCULUS. [b. V. CH. Till.
Sum of any two perpendicular radii of curvature. •
1C4. Corollary. The sum of (950 and 951) is
11^^ [\ +DyZ')Dlz-^D.z D^ zDl,,^z^{ \+D^z^)Dlz
' ^ ' ""'^' (952)
which is, by (918), the sum of the reciprocals of the
greatest and least radii of curvature at the point x^ y, z ;
or it is the sum of any two perpendicular radii of cur-
vature.
165. Problem. To find the greatest or least surface
which can he drawn under given conditions.
Solution. This form of statement embraces that portion of
the problem of § 146 which was reserved for this chapter.
Since a single equation between the coordinates of each point
is sufficient to determine the surface, no such equation can be
given ; but there may be particular conditions invohing defi-
nite integrals, like those referred to in § J 46.
166. Corollary. When there is no condition what-
ever, the required surface is absolutely the least surface
of all lohich have the same boundary.
In this case, the integral to be a minimum is (Sli or 813),
the variation of vrliich gives
ff^. COS.., {D,zDJz+DyzD^dz)=i^. (953)
But, by integration,
fz fy . COS. y Dxz D-^sz =fyfz . cos. y Dj,Z D^Sz
=/y . Dz z cos. y 8 z—frfy . D:c (cos. Y Dj^z) s z, (954)
(955)
ffj . cos.y Dy zDy ^ z—fj. .By z COS./' (J z-fxfy . By (cos. yBy zy z ;
<§> 167.] CURVATURE OF SURFACES. 193
Least surface.
whence, by regarding only the terms under the double sign of
integration,
0=zDj: (cos. y D^ z) + Dy (cos. •/ D,j z)
=zcos.y{D%z+D'yz)+Dx z Dz . cos. y-VDy z Dy cos. /. (956)
But
Dz. COS. y=zDz. {DzZ^ + DyZ'^+X)"^
= — C0S.3 y {DzzDlz + DyzDl.y z), (957)
Dy. cos. Y -^ — COS? Y{DzzDl,yZ+DyzDlz) ; (958)
which, substituted in (956), give by (952),
^^ (1 +D, %^) Dlz—2 DzzDyz Dl.yZ + jX+D^z-^) Dlz
sec.3 y
or 5' = — (?; (960)
so that this surface is one in which every point is a case
of <§) 158 ; that is, i?i which the curvatures, in directions
'perpendicular to each other , are equal and opposite.
The plane is the most simple instance of such a sur-
face, but there are other examples to an milimited ex-
tent.
167. Corollary. The complete determination of these sur-
faces must be reserved for a chapter upon the integration of
partial differential equations ; but the following ingenious con-
struction, proposed by Monge, notwithstanding its obvious want
of practical utility, which was acknowledged by its author, is
17
194 INTEGRAL CALCULUS. [b. V. CH. VIII.
Construction of minimum surface.
sufficient to exhibit the possibility of such a surface, and give
some idea of its nature.
Let any curve line, of single or double curvature, be drawn
at pleasure in space. Produce all its radii of curvature
towards the opposite side of the curve from the centres of cur-
vature, and to a distance from the curve exactly equal to the
corresponding radii of curvature. The given curve line may,
then, be assumed as a line of curvature of the required sur-
face ; that is, as a line which lies upon the surface and has at
each point, the same curvature with the surface in the direction
of this line. The produced radii of curvature, will be the radii
of curvature of the surface in directions perpendicular to the
given curve ; and if the extremities of those produced radii,
which are the corresponding centres of curvature, are fixed,
and if all the points of the given curve are rotated with the
radii about these centres, moving in planes perpendicular to
the given line, each element of the given line will describe an
element of the required surface. The given line in its new
position will acquire a new form and become a new line of cur-
vature, from which another elementary zone of the surface may
be described by a repetition of the above process.
The small arc, through which each point of the curve must
move, is not arbitrary, but is limited by the condition that two
successive radii must be in the same plane, so as to meet at
the centre of curvature.
168. Corollary. If the given curve of the preceding con-
struction were a circle, the resulting surface would be a sur-
face of revolution about an axis perpendicular to the plane of
the circle and passing through its centre. The particular form
<5> 168.] CURVATURE OF SURFACES. 195
Minimum surface of revolution.
of this surface may be investigated by taking the axis of z for
that of revolution, so that if
,, =:^2 + y2^ (961)
z will be a function of u^ and will contain no other function of
X and 7j. Hence
D^zz=zD^z.D^u — ^xD,z, ']
^ ^ ' K (962)
which, substituted in (958), give, by dividing by 4 cos.3 y,
2>„ 2 + 2 ^^ B^z"^ ^uDlz= 0. (963)
By putting i; = a/ w, (964)
we have ^ 1 r^
D^z — —- D,z,
Z V
B^ z— B ^-1- —~B^z'
(965)
which, substituted in (963), give
B z A- B z^
— [^ +D;z= 0. (966)
Hence 1 Biz _
^ = v+B^z+B^z^-^
\ . Biz B^zBlz
the integral of which is, by introducing A as an arbitrary con-
stant,
log. ^=log. I'+log. B, z— log. V (1+A ^2), (968)
196 INTEGRAL CALCULUS. [b. V. CH. VIII.
Minimum surface of revolution.
or A D,. z
V V{i+D,z2)'
(969)
Hence _ A
and if q is taken so that
t;=^Cos. y, (971)
(970) gives
Dip . z=: D^z . Dip V =1 A Sin. cp D^z
= ^Sin. ^ . -r^^— = A, (972)
z= Acp; (973)
and the equation of the surface is
iAieUe-^)- (^'^)
§ 169.] CUBATURE OF SOLIDS. 197
General expression for the element of volume.
CHAPTER IX.
THE CUBATURE OF SOLIDS.
169. Problem, To find the measure of the volume
of a given solid.
Solution. Let the conditions of the bounding line be ex-
pressed by an equation between three variables, /, m, and n.
Suppose two surfaces drawn infinitely near each other, in such
a way that n is constant throughout their extent. If, then, V
denotes the required volume, we have
€?„ F= the lamina included between these two surfaces.
If two other surfaces are drawn infinitely near each other, in
such a way that m is constant through their extent, we have
djn dn V=z the small solid rod included between these four
surfaces.
If tw^o more surfaces are drawn infinitely near each other, in
such a way that / is constant throughout their extent, we have
di djn dn V =: the infinitely small parallelopiped included be-
tween these six surfaces. (9T5)
If s' denotes an arc of the intersection of two surfaces for
which 7n and ?i are constant, s" an arc of the intersection of
two surfaces for which / and n are constant, s'" an arc of the
intersection of two surfaces for which / and /« are constant;
17*
198 INTEGRAL CALCULUS. [b. V. CH. IX.
General expression for the element of volume.
and if a! is the inclination of s" to s'" at the point of meeting,
a!' that of s' to s'" ^ and a!" that of s' to s" ; and if h is the in-
clination of s'" to the surface which includes s' and 5" ; the
sides of the small parallelopiped will be c?s', ds" y ds'" ;
the face which includes d s' and d s"^= sin a'" d s' d s"
the distance of this from the opposite face = sin. h ds'" \
whence
di d^ d, F= sin. a'" sin. h ds' d s" ds'". (976)
But I is the only variable in 5', m the only one in 5", and n the
only one in s'" , whence the accents may be neglected, by di-
viding by dl . dm . dn f and (976) gives
(977)
Dl^,r^ V=D,,D^ . Z>„ V- sin. a'" sin. b D, s . D,,s . D,,s ;
in which DiS, D„^ s, and D„ s may be deduced from the gen-
eral expression for the differential of an arc in space, by put-
ting successively each pair of the quantities /, m and w, equal to
zero. The value of V is, then, the third integral of (977).
170. Corollary. If one of the vertices of the parallelopiped
is taken for the centre of a sphere, a\ a", a'" will form, by the
intersection of the sides of the parallelopiped with the surface
of the sphere, a spherical triangle ; in which h will be the dis-
tance of a'" from the opposite vertex.
Hence, if A' is the angle opposite a', and if M is the ratio
of the sines of the sides to the sines of the opposite angles, so
that
sin. A
31=-. p, 978)
sm. a' ^ '
we have
sin. h =r sin. a' sin. ^'= iHf sin. a! sin. o!' ; (979)
§ 173.] CUBATURE OF SOLIDS. 199
Cubature of solids of revolution.
and (977) becomes
Z>f.„.„.F=^-^sm.a'sin. «"sin. d^.D.s.D^s . D^s. (980)
171. Corollary. If I, m, n are the rectangular coordinates
z, y, z, we have by (725),
ds^- = cl x^ + chf- + dz^ (981)
a' = a" = a'" = ^^, 31 z= i ; (982)
and (980) gives
Dl.,.^.V=l, (983)
V=f.fJ. 1 =U, z -/./. y =fyf. ^. (984)
172. Corollary. If /, m, n are the polar coordinates of
§ 73 of B. I., the equations (31, 32, 33) of vol. 1 give, by put-
ting
u :=. r sin. ip '\
y z=i u cos. 6 > (985)
z z=z u sin. <5j )
dy'^-\-dz^—d ifi + w2 ^^2 (9S6)
= f/r2+r2 rf52_|_^2sin.2^,rf^'2, (987)
D^ ^ ^F=r2 sin. c/^, (988)
V=fl^^r^^^^n.,=-fl ^ ^.2eos,,=/^^r3sin.,.(9S9)
173. Corollary. If the coordinates are 2:, ?<, 5 of the pre-
ceding corollary, (987) gives
Dl.u.^V=u (990)
200 INTEGRAL CALCULUS. [b. V. CH. IX.
Volume of spliere, ellipsoid.
174. Corollary. If the given solid is one of revolution about
the axis z, of which a segment is required formed by two planes
perpendicular to the axis of revolution, z may be substituted
for X in (991), and the integrals relative to 6 taken from to
2 -^. Hence
V—'XnJ^^ ^.u'=nf^.u'^ = 2nr.zu. (992)
175. Examples.
1. To find the volume of the segment of a sphere.
Solution. If Jt is the radius of the sphere, and if the axis
of z is perpendicular to the bases of the segment, (992) gives
= ^RHz,-z,) — in(^zl^zl). (993)
Corollary. The solidity of the sphere is fyr/ja^ (994)
2. Given the volume Fof a solid included within any sur-
faces whatever, the combination of which, considered as one
surface which in general is discontinuous, is represented by
the equation
i^. (a: . 3/ . z) = 0, (995)
to find the volume V of a solid included within the system of
surfaces
^.(|>.l) = 0. (996)
Ans. V'=ahc V. (997)
3. To find the volume of the segment of an ellipsoid in-
cluded between two planes drawn perpendicular to either of
the axes of the ellipsoid.
Ǥ> 175.] CUBATURE OF SOLIDS. 201
Volume of hyperboloid and paraboloid.
Ans. U A, By C are the axes of the ellipsoid, if the planes
are drawn perpendicular to the axis of C, and if V is the
solidity of the segment of a sphere whose radius is unity, the
segment being included between two planes drawn at the dis-
tances -^ and -^ from the centre, the required volume is
F'= ABCV. (998)
4. To find the volume of the segment of an hyperboloid in-
cluded between two planes drawn perpendicular to that axis,
for which the sections made by the planes are elliptical.
Ans. If Cis the axis perpendicular to the planes, and if ^
and B are the other two axes, the required volume is
V=i^^ {zl-zl) ^nAB {z-z,), (999)
in which the upper sign corresponds to the hyperboloid of one
branch, and the lower sign to the hyperboloid of two branches.
5. To find the volume of the segment of the paraboloid, in-
cluded between two planes drawn perpendicular to the axis of
z, the equation of the paraboloid being
(-.)■+ (!)'= <^- <■■>
Ans. ^n ABC{z\ — zl). (2 a)
6. To find the volume of the segment of a solid of revolu-
tion included between two planes, drawn perpendicular to the
axis of revolution, when the revolving arc is that of a cycloid
about the axis of x in (130) of vol. 1.
Ans, Vz=i I jR2 jt (sin. 2 fl,— sin. 2 ^^o)— 2 R- ^ (sin. ^,— sin. ^o)
+ 3i22-(^-^o). (3a)
202 INTEGRAL CALCULUS. [b. V. CH. IX.
Solid of least surface.
7. To find the volume of the segment of the solid of revo-
lution of § 174, when
u = B Cos. J. (4 a)
Aus, V=iAB'-n(Sm!^-^'-Shi^-^\-\-iB^7v{z^-z,). (5a)
176. Prohlein. To find the inaximuin or ininimiini
volume which can he included by a surface drawn under
given conditions.
Solution. Since the general expression for the volume is
reduced to the form of a double integral, this problem is pre-
cisely similar in its solution to that of § 165.
177. Examples.
1. To find the maximum or minimum volume^ which
can be included within a surface of a given area.
Solution. Since the double integral (984) is to be a maxi-
mum, while that of (811) is to be constant. We have, by
§ 166, if ^ is a constant multiplier,
or 1 I 1 ^
+ '- = -i; (7 a)
that is, the surf ace is one for which the sum of the recip-
rocals of the greatest and least radii of curvature at each
<5> 177.] CUBATURE OF SOLIDS. 203
Solid of revolution of least surface.
point is constant. The general equation of this sur-
face has never been obtained, but the sphere and the
cylinder are evidently cases of it.
2. To find the solids of revolution which are solutions of the
preceding problem.
Solution. Let the axis of z be that of revolution, and by
putting
u = ^(x^+y^), (8 a)
(7 a) becomes, by means of (952),
Let V be taken so that
uD,
V
(10 a)
whence
log. V = log. u + log. D,,z — i log. (I + Z), z2) ; ( 11 a)
the differential of which is
D^v I . Dlz D^z Dl z
— =--rTr
V ~~ u ' D,z l+D^z^
1. , mz
u '^ Dl{\+D,,z')
= ^+7^o4rV7r^; (12a)
which, multiplied by (10 a), gives by (9 a),
(13 a)
The integral of this equation is
u-'
v = -^-^ + B, (14a)
204 INTEGRAL CALCULUS. [b. V. CH. IX.
Solid of revolution of least surface.
in which JB is an arbitrary constant. But if r is taken so that
D„ z := cot. T, (15 a
(10 a and 14 a) give
V =: u COS. T, (16 a
COS.r=: — -^ + - , (Ha
2 A u
V(-^ + cos.3^)=-+-, (18a
u =Acos.r-/y{2AB+A''cos.\), (19 a
^1 ^^sin. tCOs. T
^ ' /s/(2AB+A^cos.Ty ^
-A ^ cos ^ "^
^ ' //(2^S+^^cos.^t) ^
If e is taken so that
'= ^{A2 + 2ABY' (^^^
(21 a) gives, by the notation of elliptic integrals,
\/(2 A B+A^ cos, 2 r)— a/(^2_j_2 AB) ,^r, (23 a
, A^—2AB , 2^^
i>^.=.-^ COS. ^+;7 ^^ V, ^^^, , x + V(Z^^-^)^^- ('''
z = — ^ sin. T -|- (yl_2 B)eF'^ + 2eBEr ; (25 a
and z may be found in in terms of u, by substituting (19 a) in
(25 a).
The preceding solution applies strictly to that case only in
which A and B have the same sign ; for, when they have op-
posite signs, e becomes greater than unity, and when B is also
greater than A, e is imaginary ; but these cases are solved
without difficulty.
<§» 177.] CUBATURE OF SOLIDS. 205
Greatest solid under given conditions.
3. To find the greatest solid of all those for which
/./.sec.2y, (26 a)
has a given value, 7 being the inclination of the tangent plane
of the bounding surface to the given plane of 2?/.
Solution. If ^ is the constant multiplier of (26 a), the equa-
tion of the maximum is
l — A-'Dlz — A-'D;z=0, (27 a)
which is easily derived from the equation
sec.2 v = l + D, z"^ + /), z\ (28 a)
Let V be taken so that
z=:lA{x + yYJ^v, (29a)
which gives
Dlz^^A+Dlv, (30 a)
Dl% = iA + Dlv- (31a)
which, substituted in (27 a), give
Dlv+Dlvz=:0. (32 a)
Let now
mz^ x-\- y \/ — 1 , w — X — y s/ — 1 ; (33 a)
and w^e have
DyV ziz {D„,v — D,v)\/\, W34a)
Dlv = Dlv\-<^ Dl,„v + Dlv,
/>> = — Z>> + 2 J9L. V — Dl V ; J
which, substituted in (32 a), give
r —Dm.D, V. (35 a)
>2
m . n
18
206 INTEGRAL CALCULUS. [b. V. CH. IX,
Greatest solid under given conditions.
Hence D„ i? is a function, whose difTerential coefficient taken
relatively to m is constant, and may, therefore, be any function
whatever of n, represented by N ; that is,
Dr. V = N, (36 a)
Hence v =f, N —f.71 + F. nij (37 a)
in whichy and JP are any arbitrary functions ; f . n is the
function whose differential coefficient is N, and F.ni is the
arbitrary quantity which is constant relatively to n ; that is,
which does not vary with n, but may be any function whatever
of the other variable m, and which is added to complete the
integral. By the substitution of (33 a), (37 a) gives
v=f.{x + ys/-\)+F[x-y^^\). (38 a)
If we put F — f'^^—i,Fi _ (39 a)
F^f'-- ^—l,F\ (40 a)
in which f and F' are real functions, the value of y becomes
(41 a)
»=/.(x+V-l)+/.(a:-^-l)+v'-l[-F'(^+'«/-l)--F'(^v-I)],
from which the imaginary quantities will wholly disappear.
<§> 180.] LINEAR DIFFERENTIAL EQUATIONS. 207
Order and degree of differential equations.
CHAPTER X.
INTEGRATION OF LINEAR DIFFERENTIAL EQUATIONS.
178. A differential equation is said to be of the same
oi'de?' with that of the highest differential coefficient
which it involves.
The degree of a differential equation is determined
in the same way as that of an ordinary equation, except
that the independent variables are neglected, and each
differential coefficient is counted as a variable.
Thus the equation
A Dlv + B D"-^ V + &z.c.-\- A'Dlv -\- B' D-' v + &c.
-\-E Dl-"". D';; V 4- &c. +ez; -{-n=0. (42a)
is of the n order ; but it is only of the first degree, or linear,
if the coefficients A, B, &,c. involve the independent variables
X, y, &:,c., but do not involve v, &/C.
179. Any equation, which is of a less order than a
given differential equation, and satisfies it by the aid
of differentiation without the assistance of any other
equation, is said to be an integral of the given equation.
The integral is said to be complete when it contains the
greatest possible number of arbitrary quantities.
180. Problem. To integrate several given equations^
between the variables x, y, z, i^'c, and their differential
coefficients taken with respect to the independent varia-
208 INTEGRAL CALCULUS. [b. V. CH. X.
Linear diHerential equations with constant coeliicienls.
hie tj xolien the given equations are linear^ and contain
no term independent of x, y, z, t^'c, and wJien all the
coefficients are constant, and the number of equations
the same with that of x, y, z, ^*c.
Solution. If the following expressions are assumed for the
variables
x = A c'\ y — Be'' &c., (43 a)
in which 5, A, B, &/C., are constant, their differentials give
DtX=i As e'\ Dty— Bs e'', &lc. )
D]x=As^e"-, D\y — Bs^e'\^(i. C (44a)
&c. &c. 3
If these values are substituted in the given equations, these
equations will evidently become divisible bye*'; and the di-
vision by this factor will free the equations wholly from
variables, and reduce them to equations between s, A^B, &c,,
in which A and B will have a linear form. If all of the con-
stants A^ B, &LC. but one, as A, are eliminated, the result will
be a single equation involving A and s, in which A, however,
will be a factor of the whole equation ; so that the division of
this equation by A, will lead to a final equation, involving no
other unknown quantity but s, and which will serve to deter-
mine s. Let the equation for determining 5 be denoted by
^ = 0, (45 a)
and each root of it will give corresponding values of A, B,
&/C., or rather of their ratios, and thence values of x, y, &>c.,
which will be integrals of the given equations.
181. Corollary. The number of integrals found by the
preceding process, will be the same as that of the different
roots of the equation (45 a) ; but all these integrals can be
<5» 183.] LINEAR DIFFERENTIAL EQUATIONS. 209
Linear differential equations with constant coefficients.
united into one expression. For it is evident that, if x^ , y^ ,
&c. denote any one of these systems of integrals,
x^z Lx,-\-L' x,,-\-^z.y y=^Ly^-[- L' Us' -{• ^C" (46 a)
will also be a system of integrals, in which L, i', 6lq,. will
be arbitrary ; for the linear form of the given equations will
cause the multipliers of L, L\ &,c. to become th esame func-
tions of x^ , y^ , &,c., which the whole equations are of x, y,
&,c. ; and therefore x,, , y^ , &/C, will satisfy the equations in
the same way as they do when they are by themselves ; that is,
the aggregate of the terms dependent upon them will be zero.
182. Corollary. If the first member of the equation (45 a)
is reduced to the form
s" + « s"-^ + &c., (47 a)
the expressions
will, by the notation of the residual calculus, include all the
terms of (46 a), provided that the residuation is performed
relatively to s, and that Aj B, C, &lc. assume a new system of
values for each root of -2A^.-D.V'D^,V+(l+(AV)2)D^Vr=0.
(81 e)
By the substitution of the preceding example this equation be-
comes
(l+p2) Dlcp + 2 p s Dl. (p + (I + s^) D', 9=0, (82 e)
4> 228.] DIFFERENTIAL EQUATIONS OF SECOND ORDER. 285
Integration of equation of surface of minimum extent.
which cannot be integrated by the direct application of the
general process. If^ however, we put
cp' = D,cp, (83 e)
the differential coefficient of (82 e), relatively to s, is
(1 +p^) Dl (p' + 2 ps Dl^ip' + (1 + s2) Dl ^) - 0, (86 e)
the integral of which, found by the process of Ex. 27 of
§220, is
.-S;--^C--(f^')'].'->
in which 5^ and p should be accented when the lower sign is
used.
Instead of proceeding with the direct process, we may put
,n=P^^^, n=P^, (88e)
5 5^ S S^
T T
and (87 e) gives
^ = ws + V (— 1— w^) = 7is— V(— 1— «^), (89 e)
(l+s2)m2— 2psm + (l+p2) = |
(l+s2)n2 — 2psw+(l+/?2)=0, ) (^^®)
Dr,p =zm D^s, D^p = n D^s, (92e)
286 INTEGRAL CALCULUS. [b. V. CH. Xll.
Integration of equation of surface of minimum extent.
D,, cp' = {m Dp cp' + D, L. s
= — [(I +p^)Dlcp' -\.2ps Dip >' + (1 + s2) Dl cp']
l±f (2pDpCp'-{-2s D,cp'){l +5^)
4 V (-l-w^) V (-1-71^) 4 V (— 1 — wi"2} V ( — 1 — ^2)
= 0. (94 e)
Hence we find by integration,
cp' =z F.m — F.m -^ -\-f. n, (95 e)
in which f. n is the function which , X. h\ (4 f )
But if ?7i is less than unity, A"*" will be infinitely greater than
A", and the equation (4 f ) becomes
P' X"^=- 0, (5f)
which is impossible, so that in this case (4 f ) cannot be satis-
fied, and (97 e) is not a case of (98 e), and is consequently a
particular solution.
If »» had been unity, (4 f ) would have been reduced to
P' X' = D, X', (6 f )
which is easily satisfied.
If m were greater than unity, (4 f ) becomes
D, X' = 0, X' = constant, (7 f )
so that a particular solution is only indicated by the condition
that m is less than unity.
The differential of (2 f ) gives
X>,, p' =.mP' {x' — xY-\ (8 f )
which, when x' differs infinitely little from x and m is less than
unity, gives
D.p^-'^i; (9f)
that is, DxP is a fraction whose denominator is zero.
<5. 232.] PARTICULAR SOLUTIONS. 289
Particular solutions.
The diflferentialion of (96 e) relatively to x gives, by substi-
tuting J) for D^ X,
D^ R . D^p + i>, 72 =: 0, (10 f )
''■^ = -d;r- (i»o
Whence by (9 f ),
i>^R=0, (12 f)
provided the numerator cannot become infinity, which will
be the case when (96 e) is free from radicals and fractions.
This equation (12 f) corresponds to the particular solution,
and leads to the particular solution by the elimination of p be-
tween it and the given equation (96 e).
231. Corollary. A similar method of finding particular so-
lutions may be extended to other differential equations.
232. Examples.
1. Find the particular solution of the equation
t + xD^x — d,x ^ {x^-\-t^ — a^). (13 f )
Solution. This equation, freed from radicals, becomes
whence (12 f) becomes
X {t + xp) =ip (x^ + i^ — «-)•
The elimination of /; gives for equation
(x^ — a^) (x2 4-^2_a2) = 0,
of which the factor
x2 _!_ ^2 _ ^2 — (14 f)
is the particular solution.
25
290 INTEGRAL CALCULUS. [b. V. CH. XlfL
Particular solutions.
2. Find the particular solution of the equation
x — tD,x-\-P, (15 f)
in which P is a given function of Z>, x.
Ans. It is the equation obtained by the elimination of 2?
between the equation
^^ < (I6f)
and t -\- DpP' — 0, ^ ^ '
in which P' is the value of P obtained by the substitution of
p for jD^ X.
THE END.
/•/ (-52 a)
<§> 186.] LINEAR DIFFERENTIAL EQUATIONS. 211
Integration of linear differential equations with constant coefficients.
186. Corollary. If x^ , ij ^ , (fcc. ; Jo, ?/o , &lc. ; z'J , &/C.,
represent the values of cT, y, &/C. ; D^ x, D^ y, &/C. ; D]x^ &/C.
when ^ vanishes; and if
/
(53 a)
equations (51a and 52 a) give
If the number of the equations (54 a) is taken equal to that of
the constant «, ,'?, &c., the values of «, (5, &,c., may, by the
usual process of elimination, be found in terms of Xq , i/^ , &:,c.
The expressions of «, ^, &;c. in terms of x^ , i/^ , &;c. will
clearly be linear functions of «, (^, &c. ; so that if these values
are substituted in (51 a and 52 a), the expressions of t, y, &/C.
will contain x^ , y^ , &c., in the same linear form in which
they now contain «, ,^, &/C. The values of «, i^, &lc., in (51 a
and 52 a), might, then, have been assumed at once as identical
with Xq , ?/q , &c., and the corresponding values of x, ?/, &c.
would be
J. Zz a;^ + Z; x'o + &c. + iJ/x 1/q + &c.
X =z l^ " — . , ^. X — ^ e'
&LC,. ;
(55 a)
s£
2' - C ((Sj)
in which, it may be observed, that the values of Z/j, Mx, &/C.
are entirely distinct from those in (51a and 52 a).
212 INTEGRAL CALCULUS. [b. V. CH. X.
Residual integral of a rational fraction.
187. Lemma. If JP denotes the value which xf.x
acquires when x becomes infinite, we have
F=l{(f.x)), (56 a)
whenever /. x denotes a rational fraction, of which the
degree of the numerator is less than that of the de-
nominator.
Proof. It follows from (216 and 219), that, in the present
case.
the product of which by x is
.f.. = lp^\ , (58 a)
But when x is infinite, (58 a) becomes
F=l{{f.z))=t{{f.^)). (59 a)
188. Corollary. When the excess of the degree of the de-
nominator of /. x above the numerator is greater than unity,
(59 a) becomes
0=t((/-^)). (60 a)
189. Corollary. When the excess of the degree of the de-
nominator of y. X above the numerator is exactly unity, and
when f ,x is of the value (217), (59 a) becomes
a
'a!
-,^L{U'A)' (61a)
190. Corollary. Since, when t becomes zero, the values of
X, y, &/C. (55 a) are reduced to x^ , y^ , &,c. ; the polyno-
<§) 190.] LINEAR DIFFERENTIAL EQUATIONS. 213
Integration of linear differential equations with constant coefficients.
mials L^ , 31^ , Ly , L'y , &c. must be of a less degree than
the (n — l)th; while Lx , My , &c. must be of the form
5"-i 4- J s"-2 + &c. (G2a)
The form of 5 L^ is, therefore,
s" + b s"-' -{- &LC. ; (63 a)
so that, by (47 a), s i^ — S (64 a)
is of a less degree than S. We have, then, by denoting (64 a)
But when t vanishes, Dt x is reduced to Jq > and therefore
s Lt'x must be of the form (62 a), while L,^, , s Mz , &lc. must
be of a smaller degree. We have then, again, by the differen-
tiation of (65 a),
and a similar train of argument may be continued to the higher
differential coefficients.
191. Corollary. If, in the given equations, there are
substituted for x, D^ x, D] x, &,c., the quantities contained
under the sign of residuation in (55 a, Go a, 6G a, &lc.), those
equations must be satisfied. The reverse process, therefore,
of substituting for DtX, D^y, &:-c., not 5 z, sy, &,c., but
sx — 2;q>S^.— , sy — .yo^-^,