1
Why we cannot divide by zero
Michael J. Neely
University of Southern California
[Link]
These notes discuss why we cannot divide by 0. The short answer is that 0 has no multiplicative inverse, and any attempt
to define a real number as the multiplicative inverse of 0 would result in the contradiction 0 = 1. Some people find these
points to be confusing. These notes may be useful for anyone with questions about dividing by 0. In order to ask why the
division operation 1/0 is not defined, it is important to understand where division comes from in the first place. Thus, the first
part of these notes discuss axioms of arithmetic. Readers looking for quick insight and intuition, without all the mathematical
overhead, can skip to the brief sections D, E, F .
A. Axioms
Arithmetic starts by assuming there are objects called real numbers. It assumes these real numbers can be manipulated by
operations of addition and multiplication that take two real numbers and produce a third:
• Addition: For any two real numbers a and b, there is a real number a + b.
• Multiplication: For any two real numbers a and b, there is a real number ab.
These addition and multiplication operations are assumed to satisfy certain axioms, called axioms for a field. The axioms
include standard commutative, associative, and distributive laws for arithmetic. The complete list of axioms is below:1
1) Commutative law: a + b = b + a and ab = ba.
2) Associative law: (a + b) + c = a + (b + c) and (ab)c = a(bc).
3) Distributive law: a(b + c) = ab + ac.
4) Existence of 0: There is a number “0,” called the additive identity, that satisfies a + 0 = a for all real numbers a.
5) Existence of 1: There is a number “1,” called the multiplicative identity, that satisfies a1 = a for all real numbers a.
6) Additive inverse: For every real number a, there is a real number −a, called the additive inverse of a, that satisfies
a + −a = 0.
7) Multiplicative inverse: For every nonzero real number a, there is a real number a−1 , called the multiplicative inverse of
a, that satisfies aa−1 = 1.
8) 0 6= 1.
The axioms cannot be proven: They are a short list of properties that we intuitively expect numbers to satisfy. However, the
axioms are useful in that, from them, important additional facts about numbers can be proven. An important theorem that can
be proven directly from the axioms is that multiplying any real number by 0 produces 0 as the result.
Theorem 1: For every real number x, the following equation is true:
0x = 0
Proof: See Appendix A for the standard proof.
B. Criticizing the multiplicative inverse axiom
One immediately notices an asymmetry in the way the additive and multiplicative inverse axioms above are stated (axioms
6 and 7 in the above list). The additive inverse axiom holds for all real numbers, while the multiplicative inverse axiom holds
only for all nonzero real numbers. Why is 0 excluded? This is not fair! For a direct answer to this question, the reader can
skip to Section D. Readers who prefer the question “why can’t we divide by zero?” must admit that this question cannot even
be asked until we define division, which is done in the next section.
C. Subtraction and division defined as inverse operations
The axioms are stated in terms of addition and multiplication operations. Operations of subtraction and division are defined
after the axioms are stated. Here are the standard definitions.
• Subtraction: For two real numbers a, b, the subtraction a − b is defined as a + −b.
−1
• Division: For two real numbers a, b, with b nonzero, the division a/b is defined as ab .
In particular, 1/2 is defined as 2 and 1/5 is defined as 5 . Thus, 1/0 is not defined, simply because there is no 0−1 .
−1 −1
That is, 0 is not defined to have a multiplicative inverse.
1 The 8th axiom “0 6= 1” can equivalently be replaced by the assumption “there exists a nonzero real number.” Neither of these statements can be proven
without the other, since the case of all numbers being 0 is consistent with the 7 other axioms. Strictly speaking, the real numbers require more structural
assumptions to distinguish them from other fields that satisfy the 8 axioms. That additional structure is not relevant to these notes and hence is omitted.
2
ad+bc
As an example, if bd 6= 0 then the expression bd is defined as:
ad + bc
= (ad + bc)(bd)−1
bd
This is the multiplication of the “numerator” (ad + bc) and the multiplicative inverse of the “denominator” bd. The value
(bd)−1 is well defined since bd is assumed to be nonzero. It can be proven from the axioms that bd 6= 0 if and only if b 6= 0
and d 6= 0. Thus, if bd 6= 0 then both b−1 and d−1 exist. This can be used to derive standard rules for manipulating fractions
such as (see Appendix B for the derivation):
a c ad + bc
+ =
b d bd
which holds if b 6= 0 and d 6= 0 (else, one or more of the expressions are not defined). Another standard rule is:
1 c
=
b/c b
which holds if b 6= 0 and c 6= 0 (else, one or more of the expressions are not defined).
D. The contradiction if 0−1 is defined
From the definition of division, the following questions are equivalent:
• Why can’t we divide by 0?
• Why is the expression a/b undefined if b = 0?
• Why is the expression 1/b undefined if b = 0?
• Why doesn’t 0 have a multiplicative inverse?
Let’s focus on the last question. Why don’t we define a particular number to be the multiplicative inverse of 0? Let’s suppose
we choose a particular real number z that we define as the multiplicative inverse of 0, so that z = 0−1 . Thus, by definition of
a multiplicative inverse:
0z = 1
However, we also know from Theorem 1 that 0z = 0. This means 1 = 0, a contradiction. Thus, we cannot assign any real
number to be 0−1 without reaching a contradiction in our pre-established facts.
E. Let’s make z a non-real number
We could define z = 0−1 to be a new symbol, not a real number. But in that case, we have to admit that this new symbol
z may not obey the rules of arithmetic that real numbers do.
Once we admit that z may not obey the rules of arithmetic, we have the complicated task of figuring out which rules it does
satisfy without leading to a contradiction. Even if we did that, we would have to figure out a reason why we are introducing
such a number z. What is the use of it?
Since there is no apparent usefulness of such a non-real number z, it is easier to simply avoid trying to define it. Then, we
just admit that 0 has no multiplicative inverse.
F. Intuition on reversibility
Here is an intuitive reason why 0 should not have a multiplicative inverse: An invertible operation should be “reversible” in
the following sense:
If I take a number x and multiply by 5, I get 5x. I can “reverse this” by multiplying by 5−1 (equivalently, dividing by 5)
to get x back. Multiplying by 5 does not lose any information about the original number x. For example, if x is unknown,
but I tell you I multiply x by 5 and the result is 10, you can determine what x was just by solving the equation 5x = 10.
The invertible operations of multiplying and dividing by 5 are like a photocopy machine that shrinks or expands a document
by a factor of 5 depending on which button is pushed. We can shrink and expand as many times as we like, with no loss of
information.
Now let’s take an unknown number x and multiply by 0. The result is 0. I cannot recover x, all I have is 0. The equation
0x = 0 does not offer any information about x. The photocopy machine reduced my document to nothing! Multiplying by 0
is a noninvertible operation! We cannot go back by “dividing by 0.”
3
G. Intuitive limit arguments
It can be shown that if n is a nonzero number, then (1/n)−1 = (n−1 )−1 = n. Now let n be a very large positive number.
We intuitively understand that 1/n approaches 0 as n is increased to ∞. Thus, we intuitively expect (1/n)−1 to approach 0−1
as n is increased to ∞ (assuming 0−1 is defined in some form). But (1/n)−1 = n, which has a limit of ∞ as n is increased
to ∞. This suggests that 0−1 = ∞.
If this argument convinces you to define 0−1 as the non-real number ∞, think again. By a similar argument it holds that
1/n approaches 0 as n is decreased to −∞ (such as when n is chosen as −10000 and then −1000000000). Thus, intuitively
one expects (1/n)−1 to approach 0−1 as n is decreased to −∞. But (1/n)−1 = n has a limit of −∞ as n approaches −∞.
This suggests that 0−1 = −∞. How can 0−1 be both ∞ and −∞?
A PPENDIX A—P ROVING T HEOREM 1
The axioms are stated with an implicit understanding of the notion of an equation such as a(b + c) = ab + ac. The equation
means that the real number a(b + c) is the same as the real number ab + ac. If an equation is “true,” then another true equation
can be obtained by adding the same thing to both sides, or by multiplying the same thing to both sides.2 This is used in the
following proof. The proof also uses notation x1 to represent the multiplication of a real number x and the number 1.
Proof: (Theorem 1) Let x be a particular real number. We want to prove that 0x = 0. We know from axiom 4 that:
1+0=1
Multiplying both sides by x gives:
x(1 + 0) = x1
Using the distributive law gives:
x1 + x0 = x1
Using the fact that x1 = x gives:
x + x0 = x
Adding −x to both sides gives:
−x + (x + x0) = −x + x
Using the associative law:
(−x + x) + x0 = −x + x
Using the fact that −x + x = 0 gives:
0 + x0 = 0
Using the fact that 0 + x0 = x0 gives:
x0 = 0
Using the commutative law x0 = 0x proves that 0x = 0.
A PPENDIX B—S TANDARD RULES OF FRACTIONS
It can be proven from the axioms that for any real numbers b and c:
• bc 6= 0 if and only if b 6= 0 and c 6= 0.
−1
• If bc 6= 0, then (bc) = b−1 c−1 .
−1
• 1 = 1.
−1
• a 6= 0 if and only if a 6= 0.
We can use the above facts to derive standard rules of fractions.
Standard rules for fractions:
1) For any real number a:
a
= (a)(1−1 ) = (a)(1) = a
1
where the first equality holds by definition of a fraction, the second holds because 1−1 = 1, and the third holds because
1 is the multiplicative identity (axiom 5).
2) Suppose a 6= 0. Then:
1
= (1)(a−1 ) = a−1
a
2 Manipulations of adding and/or multiplying an equation by the same thing can be justified from the more basic concept of substituting into an equation,
but that detail is tangential and is not discussed here.
4
where the first equality holds by definition of a fraction, and the second holds because 1 is the multiplicative identity
(axiom 5).
3) Suppose a 6= 0. Then:
a
= aa−1 = 1 (1)
a
where the first equality holds by the definition of a fraction, and the second equality holds by axiom 7.
4) Suppose b 6= 0 and d 6= 0. Then:
a c
= (ab−1 )(cd−1 ) (2)
b d
−1 −1
= (ac)(b d ) (3)
= (ac)(bd)−1 (4)
ac
= (5)
bd
where equalities (2) and (5) hold by definition of a fraction, equality (3) holds because multiplication is commutative,
and (4) holds because b−1 d−1 = (bd)−1 .
5) Suppose b 6= 0 and c 6= 0. Then:
a a
= 1 (6)
b ab c
= (7)
b c
ac
= (8)
bc
where equality (6) holds by axiom 5, equality (7) holds by rule 3, and equality (8) holds by rule 4. This proves that
multiplying the numerator and denominator of a fraction by the same nonzero number c does not change the value of the
fraction.
6) Suppose c 6= 0. Then:
a b
+ = ac−1 + bc−1 (9)
c c
−1
=
(a + b)c (10)
a+b
= (11)
c
where equality (10) holds by the distributive law (axiom 3), and equalities (9) and (11) hold by definition of a fraction.
This proves the standard rule about summing fractions with the same denominator.
7) Suppose b 6= 0 and d 6= 0. Then:
a c ad cb
+ = +
b d bd bd
ad + cb
=
bd
where the first equality uses rule 5, and the second uses rule 6.
8) If b 6= 0 and c 6= 0, then:
1 1
=
(b/c) bc−1
c
= −1
bc c
c
=
b
where the first equality holds by definition of a fraction, the second equality holds by rule 5, and the last equality holds
because c−1 c = 1.
9) If a 6= 0 then:
1 1 a
(a−1 )−1 = = = =a
a−1 1/a 1
The axioms can also be used to prove that for any real number a, we have −(−a) = a and −a = (−1)a. In particular,
1 = −(−1) = (−1)(−1). Also, (−1)−1 = −1.