HISTORY OF MATHEMATICS

MATHEMATICAL TOPIC IX
CUBIC EQUATIONS AND QUARTIC EQUATIONS

PAUL L. BAILEY

1. The Story
Various solutions for solving quadratic equations ax2 + bx + c = 0 have been
around since the time of the Babylonians. A few methods for attacking special
forms of the cubic equation ax3 +bx2 +cx+d = 0 had been investigated prior to the
discovery and development of a general solution to such equations, beginning in the
fifteenth century and continuing into the sixteenth century, A.D. This story is filled
with bizarre characters and plots twists, which is now outlined before describing
the method of solution.
The biographical material here was lifted wholesale from the MacTutor History
of Mathematics website, and then editted. Other material has been derived from
Dunham’s Journey through Genius.

1.1. Types of Cubics. Zero and negative numbers were not used in fifteenth cen-
tury Europe. Thus, cubic equations were viewed to be in different types, depending
on the degrees of the terms and there placement with respect to the equal sign.
• x3 + mx = n “cube plus cosa equals number”
• x3 + mx2 = n “cube plus squares equals number”
• x3 = mx + n “cube equals cosa plus number”
• x3 = mx2 + n “cube equals squares plus number”
Mathematical discoveries at this time were kept secret, to be used in public “de-
bates” and “contests”. For example, the method of depressing a cubic (eliminating
the square term by a linear change of variable) was discovered independently by
several people. The more difficult problem of solving the depressed cubic remained
elusive.

1.2. Luca Pacioli. In 1494, Luca Pacioli published Summa de arithmetica, geome-
tria, proportioni et proportionalita. The work gives a summary of the mathematics
known at that time although it shows little in the way of original ideas. The work
studies arithmetic, algebra, geometry and trigonometry and, despite the lack of
originality, was to provide a basis for the major progress in mathematics which
took place in Europe shortly after this time. The book admittedly borrows freely
from Euclid, Boethius, Sacrobosco, Fibonacci, et cetera.
In this book, Pacioli states that the solution of the cubic is impossible.

Date: November 21, 2008.

1
2

1.3. Scipione del Ferro. The first known mathematician to produce a general
solution to a cubic equation is Sipione del Ferro. He knew that the problem of
solving the general cubic could be reduced to solving the two cases x3 +mx = n and
x3 = mx + n, where m and n are positive numbers, and del Ferro may have solved
both cases; we do not know for certain, because his results were never published.
We know that del Ferro was appointed as a lecturer in arithmetic and geometry
at the University of Bologna in 1496 and that he retained this post for the rest of
his life. No writings of del Ferro have survived. We do know however that he kept
a notebook in which he recorded his most important discoveries. This notebook
passed to del Ferro’s son-in-law Hannibal Nave when del Ferro died in 1526.
On his deathbed, del Ferro revealed at least part of his secret, the solution to
the “cube plus cosa equals number” problem, to his student, Fior.

1.4. Niccolo Tartaglia. Niccolo Fontana, known as Tartaglia, was born in Brescia
in 1499 or 1500. His father was murdered when he was six, and plunged the family
into total poverty.
Niccolo was nearly killed as a teenager when, in 1512, the French captured his
home town and put it to the sword. The twelve year old Niccolo was dealt horrific
facial sabre wounds by a French soldier that cut his jaw and palate. He was left
for dead and even when his mother discovered that he was still alive she could
not afford to pay for any medical help. However, his mother’s tender care ensured
that the youngster did survive, but in later life Niccolo always wore a beard to
camouflage his disfiguring scars and he could only speak with difficulty, hence his
nickname Tartaglia, or stammerer.
He moved to Venice in 1534. As a lowly mathematics teacher in Venice, Tartaglia
gradually acquired a reputation as a promising mathematician by participating
successfully in a large number of debates.
Fior began to boast that he was able to solve cubics and a challenge between him
and Tartaglia was arranged in 1535. In fact Tartaglia had previously discovered how
to solve one type of cubic equation, the “cube + squares equals number” type. For
the contest between Tartaglia and Fior, each man was to submit thirty questions
for the other to solve. Fior was supremely confident that his ability to solve cubics
would be enough to defeat Tartaglia but Tartaglia submitted a variety of different
questions, exposing Fior as an, at best, mediocre mathematician. Fior, on the other
hand, offered Tartaglia thirty opportunities to solve the “cube plus cosa” problem,
since he believed that he would be unable to solve this type, as in fact had been
the case when the contest was set up. However, in the early hours of February 13,
1535, inspiration came to Tartaglia and he discovered the method to solve ’cube
equal to numbers’. Tartaglia was then able to solve all thirty of Fior’s problems in
less than two hours. As Fior had made little headway with Tartaglia’s questions, it
was obvious to all who was the winner. Tartaglia didn’t take his prize for winning
from Fior, however, the honor of winning was enough.

1.5. Girolamo Cardano. Girolamo or Hieronimo Cardano’s name was Hierony-

mus Cardanus in Latin and he is sometimes known by the English version of his
name Jerome Cardan. He was the illegitimate child of a lawyer/mathematician,
Fazio Cardano. He was a brilliant physician and mathematician who loved to gam-
ble, and generally had a fascinating, though often tragic, life. Among other travails,
he was kept out of the College of Physicians because of his illegitimate birth; his
 3

wife died young; his favorite son was executed for the murder of his wife; his other
son stole large sums of money from him; and he was jailed by the Inquisition.
In 1539,Cardan was a public lecturer of mathematics at the Piatti Foundation
in Milan, and was aware of the problem of solving cubic equations; he had taken
Pacioli at his word and assumed that, as Pacioli stated in the Suma published in
1494, solutions were impossible.
Cardan was greatly intrigued when he learned of the contest between Fior and
Tartaglia, and he immediately set to work trying to discover Tartaglia’s method for
himself, but was unsuccessful. A few years later, in 1539, he contacted Tartaglia,
through an intermediary, requesting that the method could be included in a book he
was publishing that year. Tartaglia declined this opportunity, stating his intention
to publish his formula in a book of his own that he was going to write at a later
date. Cardan, accepting this, then asked to be shown the method, promising to
keep it secret. Tartaglia, however, refused.
An incensed Cardan now wrote to Tartaglia directly, expressing his bitterness,
challenging him to a debate but, at the same time, hinting that he had been dis-
cussing Tartaglia’s brilliance with the governor of Milan, Alfonso d’Avalos, the
Marchese del Vasto, who was one of Cardan’s powerful patrons. On receipt of this
letter, Tartaglia radically revised his attitude, realizing that acquaintance with the
influential Milanese governor could be very rewarding and could provide a way out
of the modest teacher’s job he then held, and into a lucrative job at the Milanese
court. He wrote back to Cardan in friendly terms, angling for an introduction to the
Signor Marchese. Cardan was delighted at Tartaglia’s new approach, and, inviting
him to his house, assured Tartaglia that he would arrange a meeting with d’Avalos.
So, in March 1539, Tartaglia left Venice and travelled to Milan. To Tartaglia’s
dismay, the governor was temporarily absent from Milan but Cardan attended to
his guest’s every need and soon the conversation turned to the problem of cubic
equations. Tartaglia, after much persuasion, agreed to tell Cardan his method, if
Cardan would swear never to reveal it and furthermore, to only ever write it down
in code so that on his death, nobody would discover the secret from his papers.
The oath which Cardano swore is reportedly:
I swear to you, by God’s holy Gospels, and as a true man of honor, not only
never to publish your discoveries, if you teach me them, but I also promise you,
and I pledge my faith as a true Christian, to note them down in code, so that after
my death no one will be able to understand them.
Tartaglia divulged his formula in the form of a poem, to help protect the secret,
should the paper fall into the wrong hands.
By the time he had reached Venice, Tartaglia was sure he had made a mistake in
trusting Cardan and began to feel very angry that he had been induced to reveal his
secret formula. When Cardan wrote to him in a friendly manner Tartaglia rebuffed
his offer of continued friendship and mercilessly ridiculed his books on the merest
trivialities.
4

1.6. Lodovico Ferrari. Lodovico Ferrari was sent, as a teenager, to be the servant
of Cardano. However, when Cardano discovered that the boy could read and write,
he made him his assistant, and quickly learned that Ferrari was quite talented.
Ferrari became Cardano’s mathematical apprentice.
Based on Tartaglia’s formula, Cardan and Ferrari made remarkable progress
finding proofs of all cases of the cubic and, even more impressively, solving the
quartic equation. Tartaglia made no move to publish his formula despite the fact
that, by now, it had become well known that such a method existed.
One of the first problems that Cardan hit was that the formula sometimes in-
volved square roots of negative numbers even though the answer was a ’proper’
number. In August 1539 Cardan wrote to Tartaglia:
I have sent to enquire after the solution to various problems for which you have
given me no answer, one of which concerns the cube equal to an unknown plus a
number. I have certainly grasped this rule, but when the cube of one-third of the
coefficient of the unknown is greater in value than the square of one-half of the
number, then, it appears, I cannot make it fit into the equation.
Indeed Cardan gives precisely the conditions here for the formula to involve
square roots of negative numbers. Tartaglia by this time greatly regretted telling
Cardan the method and tried to confuse him with his reply (although in fact
Tartaglia, like Cardan, would not have understood the complex numbers now en-
tering into mathematics):
... and thus I say in reply that you have not mastered the true way of solving
problems of this kind, and indeed I would say that your methods are totally false.
Cardan and Ferrari travelled to Bologna in 1543 and learnt from Hannibal Nave
that it had been del Ferro, not Tartaglia, who had been the first to solve the cubic
equation. Cardan felt that although he had sworn not to reveal Tartaglia’s method
surely nothing prevented him from publishing del Ferro’s formula. In 1545 Cardan
published Artis magnae sive de regulis algebraicis liber unus, or Ars magna, as it
is more commonly known, which contained solutions to both the cubic and quartic
equations and all of the additional work he had completed on Tartaglia’s formula.
Del Ferro and Tartaglia are credited with their discoveries, as is Ferrari, and the
story written down in the text.
It is to Cardan’s credit that, although one could not expect him to understand
complex numbers, he does present the first calculation with complex numbers in
Ars Magna. Solving a particular cubic equation, he√ writes √
Dismissing mental tortures, and multiplying 5 + −15 by 5 − −15, we obtain
25 − (−15). Therefore the product is 40 ... and thus far does arithmetical subtlety
go, of which this, the extreme, is, as I have said, as subtle as it is useless.
1.7. Rapheal Bombelli. In 1572, Rapheal Bombelli wrote his book Algebra, in
which explicitly uses negative numbers and zero. Moreover, he shows how manip-
ulating complex numbers can help arrive at real solutions to cubic equations, thus
demonstrating that, although they may be subtle, they are far from useless.
 5

2. Solution of Quadratic Equations

Some version of the quadratic formula has been available to most advanced
cultures of the last three thousand years. Let us review its derivation.
A polynomial is monic if the leading coefficient is 1. Two equations are equivalent
if they have the same solution set.
Let ax2 + bx + c = 0 be a general quadratic equation; Our method of solution is
known as completing the square. First we produce an equivalent monic equation,
and then we introduce a new term to create the square of a linear polynomial:
b c
ax2 + bx + c = 0 ⇔ x2 + x = −
a a
b  b b 2 c
⇔ x2 + x + )2 = ( −
a 2a 2a a
 b 2
 b2 − 4ac
⇔ x+ = 2
2a √ 4a
b 2
b − 4ac
⇔x+ =±
2a √ 2a
−b ± b2 − 4ac
⇔x= .
2a
Notice that the fourth equation can be rewritten as
 b 2 4ac − b2
x+ + = 0.
2a 4a2
b 4ac−b2
Setting y = x + 2a and n = 4a2 , rewrite this as
y 2 + n = 0.
This is depressed quadratic equation; the degree one term has be eliminated by the
b
substitution x → y − 2a , which is known as a linear change of variables.
The discriminant of the quadratic equation is
∆ = b2 − 4ac;
this determines the number of real roots. There are three cases:
(a) if b2 − 4ac > 0, there are two real roots;
(b) if b2 − 4ac = 0, there is one real root;
(c) if b2 − 4ac < 0, there are no real roots.
We point out here that negative numbers and there square roots were not ac-
cepted as actual solutions in antiquity. To some extent, this is justified, since the
search for real solutions to such an equation was not significantly compromised
by excluding these numbers. This situation changes upon consideration of cubic
equations.
6

3. Depressing Cubic Equations

Let f (x) = ax3 + bx2 + cx + d be a general cubic polynomial; we wish to solve
f (x) = 0. If we can find one zero r of the polynomial f (x), we can divide (x − r)
into f (x) to obtain a quadratic polynomial, whose zeros can be found using the
quadratic formula.
A depressed cubic equation of the type “cube plus cosa equals number” is an
equation of the form
x3 + mx = n.
We wish to take the general cubic equation f (x) = 0 and find a different depressed
cubic equation whose solution will give us a solution to f (x) = 0.
Clearly we can divide by a to obtain a monic equation x3 + ab x2 + ac x + ad = 0;
in this way, we can assume that a = 1. We wish to find a linear change of variable
that will produce an equation which lacks the quadratic term.
To discover how to do this, substitute y − h for x and multiply:
ax3 + bx2 + cx + d = a(y − h)3 + b(y − h)2 + c(y − h) + d
= a(y 3 − hy 2 + h2 y − h3 ) + b(y 2 − 2hy + h2 ) + c(y − h) + d
= ay 3 − 3ahy 2 + 3ah2 y − h3 + by 2 − 2hy + h2 + cy − ch + d
= ay 3 + (b − 3ah)y 2 + (3ah2 − 2h)y + (d − ch).
b
Now we see that if we set h = 3a , we obtain the desired result of eliminating the
quadratic term.
Thus, to solve cubic equations, we may assume that the equation is given in the
form x3 + mx = n.
 7

4. Solving the Depressed Cubic

Consider the depressed cubic equation
x3 + mx = n.
The key idea of the solution method is to write x as a difference of two quantities,
x = t − u. We wish to find appropriate quantities t, u ∈ R such that r = t − u,
where r is a solution to this equation.
By the binomial theorem, we have
(t − u)3 = t3 − 3t2 u + 3tu2 − u3 .
Thus
t3 − u3 = (t − u)3 + 3t2 u − 3tu2
= (t − u)3 + 3tu(t − u).
Note that if x = t−u, m = 3tu, and n = t3 −u3 , this equation becomes x3 +mx = n.
This reduces the problem to solving the system of equations
(1) m = 3tu;
(2) n = t3 − u 3 .
Equation (1) gives
m
. u=
3t
Substitute this into equation (2) to obtain
m3
t3 −
= n.
27t3
Subtract n from both sides and multiply through by t3 to get
m3
t6 − nt3 − = 0.
27
Now the quadratic formula gives
q
4m3
n± n2 + 27
3
t =
r2
n n2 m3
= ± + .
2 4 27
Therefore, s r
3 n n2 m3
t= ± + .
2 4 27
Now u3 = t3 − n, so s r
3 n n2 m3
u= − ± + .
2 4 27
Finally, x = t − u, so
s r s r
3 n n2 m3 3 n n2 m3
x= ± + − − ± + .
2 4 27 2 4 27
8

Upon examining the above equations, we see two ± signs in the expression for x,
which may lead one to believe we have found four solutions to a cubic polynomial.
However, to of them are the same. √ √
Consider the solution to be in the form x = 3 a ± b − 3 −a ± b. The solution
with two minus signs equals the solution with two negative signs, because factor
out the negative sign gives
√3
√
3
√3
√
3
√
3
√
3
a − b − −a − b = (−1) −a + b − (−1) a + b = a + b − −a + b.
Memorizing this formula is unnecessary if we remember the technique. We let
x = t − u, so that x3 + mx = n becomes (t − u)3 + 3tu(t − u) = t3 − u3 .
(1) Set 3tu = m and t3 − u3 = n.
(2) Solve the first equation for u to get u = m
3t .
(3) Plug this into the second equation to get t3 = n + ( m 3
3t ) .
3 6 3 m 3
(4) Multiply by t to get t − nt − ( 3 ) =.
(5) Complete the square to get t3 = n2 ± ∆.
(6) Use u3 = t3 − n to get u3 = − n2 ± ∆;
(7) Take cube root and set x = t − u.
Example 1. (A Typical Example)
Solve x3 + 15x = 22.
Solution. Let 3tu = 15 and t3 − u3 = 22. Then u = 5t , so t3 − 125
t3 = 22. Therefore
t6 − 22t3 − 125 = 0, so by the quadratic formula,
√
22 ± 484 + 500 √
t3 = = 11 ± 246.
2
3
√
Then u = −11 ± 246, so
√ √
q q
3 3
x = 11 ± 246 − −11 ± 246.


Example 2. (Cardano’s Example)

Solve x3 + 6x = 20.
Solution. Let 3tu = 6 and t3 − u3 = 20. Then u = 2t , so t3 − 8
t3 = 20, so
t6 − 20t3 − 8 = 0. By the quadratic formula,
√
20 ± 400 + 32 √
t3 = = 10 ± 108.
2
Taking the positive value for t, applying u3 = t3 − 20, and taking the appropriate
cube roots, we have
√ √
q q
3 3
x = t − u = 10 + 108 − −10 + 108.
Note that this is a real number.
A more modern solution starts by setting f (x) = x3 + 6x − 20 and seeking
solutions to f (x) = 0. We note that f (2) = 23 + 6(2) − 20 = 0, so x = 2 is a
solution. By the Factor Theorem, (x − 2) divides f (x), and dividing we find that
f (x) = (x − 2)(x2 + 2x + 10) = (x − 2)(x − (−1 + 3i))(x − (−1 − 3i)).
 9

Note that the only real zero of f is 2; thus, we have no choice but to conclude that
√ √
q q
3 3
10 + 108 − −10 + 108 = 2.

Example 3. (Bombelli’s Example)
Solve x3 − 15x = 4.
Solution. Let 3tu = −15 and t3 − u3 = 4. Thus u = − 5t , so t3 + 125 t3 = 4. From
this, t6 − 4t3 + 125 = 0, so, taking the positive square root, we have
√
4 + 16 − 500 √ √
t3 = = 2 + 4 − 125 = 2 + −121.
2
At this
√ point, instead of asserting the irrelevance of the problem, Bombelli continues
as if −1 is a perfectly acceptable quantity. He notes that
√ √ √ √ √
(2 + −1)3 = 8 + 12 −1 − 6 − −1 = 2 + 11 −1 = 2 + −121.
√ √ √
Thus if t = 2 + −1, then t3 = 2 + −121. Continuing, we have u = −2 + −11,
so √ √
x = t − u = (2 + −121) − (−2 + −121) = 4.
One verifies that 4 is indeed a solution to the original cubic equation; thus a real
solution is attained by traversing through the realm of complex numbers. 

5. Depressing a Quartic Equation

The general quartic equation can be depressed in the same manner as the cubic.
Consider
ax4 + bx3 + cx2 + dx + e = 0.
We again want a linear change of variables that will eliminate the cubic term. Here,
b
the substitution x 7→ (x − 4a works.
Problem 1. Let
an xn + an−1 xn−1 + · · · + a1 x + a0 = 0.
an−1
Show that the substitution x 7→ (x − nan ) eliminates the term of degree (n − 1).

6. Solving the Depressed Quartic

Consider the polynomial equation
x4 + px2 + qx + r = 0.
Complete the square to obtain
(x2 + p)2 = px2 − qx − r + p2 .
Let y ∈ R and add 2y(x2 + p) + y 2 to both sides to get
(*) (x2 + p + y)2 = (p + 2y)x2 + (p2 − r + 2py + y 2 ).
The right hand side becomes a quadratic in x we can choose y so that it is a perfect
square; this is done by making the discriminant equal to zero:
q 2 − 4(p − r + 2py + y 2 ) = 0.
Rewrite this as
(q 2 − 4p3 + 4pr) + (8r − 16p2 )y − 20py 2 − 8y 3 = 0.
10

This is a cubic in y, and can be solved. With this value for y, take the square root
of both sides of (*) to obtain a quadratic in x, which can also be solved.

7. Graphs of Cubics
7.1. Backdrop. Cardano had several cases for solving cubics, placing the mono-
mials on the appropriate side of the equation to create positive coefficients. As it
turns out, the sign of the constant term plays no role in the computation, so the
main cases were:
(a) x3 = mx + n (cube equals cosa plus number)
(b) x3 + mx = n (cube plus cosa equals number)
Analytic geometry on in cartesian coordinates had not been invented at the
time, so Cardano had no idea that these two cases correspond to distinctly different
geometric interpretations for the graph of the cubic. We investigate this in modern
notation using Calculus.

7.2. The Leading Coefficient. Consider the generic cubic polynomial

f (x) = ax3 + bx2 + cx + d.
If a > 0, then limx→−∞ f (x) = −∞, and limx→∞ f (x) = ∞; if a < 0, then
limx→−∞ f (x) = ∞, and limx→∞ f (x) = −∞ Thus by the Intermediate Value
Theorem, f has at least one real zero. By the Factor Theorem, if f (r) = 0,
then f (x) = (x − r)q(x), where q is a quadratic polynomial, which we can find
by polynomial division, and thus find the other two zeros (which may be real or
complex) using the quadratic formula.
We wish to solve f (x) = 0, and we realize that if we divide through by a, the
zeros of the resulting polynomials are the same as the original. Thus, without loss
of generality, we assume a = 1.

7.3. The Square Coefficient. Consider the polynomial

f (x) = x3 + bx2 + cx + d.
Note that limx→−∞ f (x) = −∞, and limx→∞ f (x) = ∞. Next we wish to discover
the role of b in the graph of f .
By differentiation,
f 0 (x) = 3x2 + 2bx + cand f 00 (x) = 6x + 2b.
If f 00 (x) = 0, then 6x + 2b = 0, so x = − 3b . Thus f has an inflection point at
(− 3b , f (− 3b )). Also, b = 0 if and only if the inflection point of f lies on the y-axis.
So, to eliminate the inflection point of f , we shift the graph of f right by 3b . The
graph of f (x − 3b ) is the graph of f shifted so that the inflection point is on the
y-axis. If we find the zeros of f (x − 3b ), we obtain the zeros of f subtracting 3b .
It turns out that
b 1 1
f (x − ) = x3 − (b2 − 3c)x − (b3 − 3b2 + 9bc − 27d).
3 3 27
Without loss of generality, we now assume that b = 0.
 11

7.4. The Cosa Coefficient. Consider the polynomial

f (x) = x3 + cx + d.
The graph of f has an inflection point on the y-axis. We wish to identify the role
that c plays in the graph of f .
By differentiation,
f 0 (x) = 3x2 + c.
Thus f 0 (0) is the slope of the line tangent to the graph of f at its inflection point.
If c = 0, then f has a horizontal tangent at its inflection point.
If c ≥ 0, then f is increasing and has no local minimum or maximum; in this
case, f has exactly one x-intercept, and so, exactly one real zero and two complex
zeros. This is the “cube + cosa = number” case.
If c < 0, we set f 0 (x) = 0 and obtain x = ± − 3c ; thus f has a local maximum
p
p c p c
at − − 3 and a local minimum at − 3 . This is the “cube = cosa + number”
case. In this case, we sometimes have three distinct real zeros, and sometimes do
not. We wish to discover a condition determining the number of real zeros.
7.5. The Constant Coefficient. Consider the polynomial
g(x) = x3 + cx.
We have an excellent idea of its graph. It is an odd function, and thus has symmetry
about the origin, and it has an inflection
√ point at the origin. If c ≥ 0,pit is increasing,
and if c < 0, it has zeros at x = ± c with local extrema at x = ± − 3c .
If we shift this graph up by d (down if d < 0), the corresponding function is
f (x) = g(x) + d.
Assume that c < 0. Let
r c  2c r c
h=g − = − .
3 3 3
This is the height (depth) of the local maximum (minimum). If we vertically shift
the graph of g by less than h, we have three zeros; if we shift by h, we have one
single zero and a double zero; if we shift by d > h, we have a unique zero. That is:
(a) |d| < |h| ⇒ three distinct real zeros
(b) |d| = |h| ⇒ one single real zero and one double real zero
(c) |d| > |h| ⇒ a unique real zero and two complex zeros
Now
4c2 c 4c3 d2 c3
|d| < |h| ⇔ d2 < h2 − (− ) = − ⇔ + .
9 3 27 4 27
The discriminant of f is
d2 c3
D= + .
4 27
Note that if c ≥ 0, then D ≥ 0. Thus, for any c, we have
(a) D < 0 ⇒ three distinct real zeros
(b) D = 0 ⇒ one single real zero and one double real zero, or a triple zero
(c) D > 0 ⇒ a unique real zero and two complex zeros
Department of Mathematics and CSci, Southern Arkansas University
E-mail address: plbailey@[Link]