Why Is a Negative Times a Negative a Positive?

In 1831 Augustus De Morgan, a mathematician in his mid-twenties who had just taken the mathematics chair at the newly founded University of London, published a short book for schoolteachers and their pupils called On the Study and Difficulties of Mathematics. In it he poses an unremarkable problem. A father is 56 and his son is 29. When will the father be twice the age of the son? Put $x$ for the number of years from now, and the equation is $2(29+x) = 56+x$, which gives $58 + 2x = 56 + x$, so $x = -2$. De Morgan checks that this satisfies the equation, and then rejects it. The negative answer, he writes, tells us that the problem was set up wrongly. He starts again with $x$ years in the past, solves $2(29-x) = 56-x$, obtains $x = 2$, and is satisfied. Earlier in the same book he had stated flatly that $3-8$ is an impossibility, because it requires you to take from 3 more than there is in 3, which is absurd.

That is the state of one of the sharpest mathematical minds of the nineteenth century, roughly a decade before he formulated the laws that carry his name. So when a student asks why a negative times a negative is a positive and finds every answer slightly unsatisfying, the student is in respectable company. Here is the answer, stated plainly before any history or geometry: a negative times a negative is a positive because it is the only assignment that keeps the distributive law true. You do not get to choose. If you want $a(b+c) = ab + ac$ to hold for negative numbers the way it holds for positive ones, then $(-1)\cdot(-1)$ must equal $+1$, and every other sign rule follows from that one. It is a theorem, not a convention, and the proof takes three lines.

What follows is that proof, the reason it generalises to every algebraic system worth the name, the concrete models that genuinely illustrate it, and an honest account of why the everyday examples feel like a swindle.

Why the everyday examples feel like a swindle

Almost every page on this question opens with apples. Three baskets of four apples gives twelve apples: multiplication as repeated addition. Then it tries to run the same argument with negatives, and the machinery seizes. What is minus three baskets? What is a basket of minus four apples? And if you somehow have minus three baskets of minus four apples, in what sense do you now have twelve actual apples?

You do not, and the writers know it. They tend to switch, mid-paragraph, to debts or to a videotape run backwards, and a perceptive reader notices the switch and concludes, correctly, that the apple picture has been abandoned rather than extended.

The deeper reason is worth stating. Negative numbers are not counts. A count of physical objects is never negative; there is no collection with $-4$ elements in it. What negative numbers encode is direction along an axis: debt against credit, left against right, before against after, downstream against upstream. A product of two negatives therefore always requires two independent reversals happening at once, and human intuition, which handles one reversal comfortably, handles two badly. That is a fact about intuition, not about arithmetic.

So the honest structure of the answer is the reverse of what is usually offered. The argument comes first and is airtight. The pictures come second, and they are illustrations of a result already established, not evidence for it.

The proof: the distributive law leaves no choice

We assume only what everybody already accepts about positive numbers and want to extend it. The relevant facts are these: addition is associative and commutative, $0$ is the additive identity so $a + 0 = a$, every number $a$ has an additive inverse $-a$ with $a + (-a) = 0$, $1$ is the multiplicative identity so $1 \cdot a = a$, and multiplication distributes over addition, meaning

$a(b+c) = ab + ac \quad \text{and} \quad (b+c)a = ba + ca.$

The first thing to establish is that multiplying by zero gives zero. This is not an axiom; it is a consequence. Since $0 + 0 = 0$,

$a \cdot 0 = a(0+0) = a\cdot 0 + a \cdot 0.$

Now add $-(a\cdot 0)$ to both sides. The left becomes $0$; the right becomes $a \cdot 0$. Hence $a \cdot 0 = 0$ for every $a$, including negative $a$, and we never had to assume anything about signs to get it.

Now the main argument. Consider the expression $(-1)\cdot\big(1 + (-1)\big)$. The bracket is $0$, so by what we just proved the whole thing is $0$. But we can also expand it with the distributive law:

$(-1)\cdot\big(1 + (-1)\big) = (-1)\cdot 1 + (-1)\cdot(-1).$

The first term is $(-1)\cdot 1 = -1$, since $1$ is the multiplicative identity. So

$0 = -1 + (-1)\cdot(-1).$

Add $1$ to both sides:

$(-1)\cdot(-1) = 1.$

That is the whole proof. Notice what it does not do. It does not appeal to any interpretation of negative numbers, any picture, any physical situation, or any appeal to symmetry or elegance. It uses the distributive law and the definition of additive inverses, and nothing else.

The general case follows by the same route. For any $a$ and $b$, first show $(-a)b = -(ab)$:

$ab + (-a)b = (a + (-a))b = 0 \cdot b = 0,$

so $(-a)b$ is precisely the additive inverse of $ab$. Then

$(-a)(-b) + \big(-(ab)\big) = (-a)(-b) + (-a)(b) = (-a)\big((-b) + b\big) = (-a)\cdot 0 = 0,$

which says that $(-a)(-b)$ is the additive inverse of $-(ab)$, that is, $(-a)(-b) = ab$.

What would break if the rule were different

It is one thing to say the rule is forced. It is more convincing to watch the alternative collapse. Suppose someone insists that $(-3)\cdot(-4) = -12$, on the grounds that two negatives ought to make something more negative. We already know that $(-3)\cdot 4 = -12$, since that is four copies of $-3$ and nobody disputes it.

Now compute $(-3)\cdot\big((-4) + 4\big)$ in two ways.

Directly, the bracket is $0$, so the answer is $0$.

By distributing, the answer is $(-3)(-4) + (-3)(4)$. Under the proposed rule this is $-12 + (-12) = -24$.

So the proposed rule yields $0 = -24$. The distributive law has failed, and with it goes essentially all of algebra: you can no longer expand brackets, factorise, multiply polynomials, or solve a linear equation reliably. The cost of making the sign rule feel intuitive is that arithmetic stops working.

A shorter version of the same collapse: if $(-1)(-1) = -1$, then $(-1)(-1) = (-1)(1)$, and cancelling the common factor $-1$ from both sides gives $-1 = 1$.

It is not a rule someone made up: the ring axioms

The proof above used exactly five ingredients: an additive group structure, a multiplicative identity, and distributivity. That list is the definition of a ring with identity. It follows immediately that in any ring whatsoever, $(-a)(-b) = ab$. The sign rule is not a property of the integers specifically; it is a structural consequence of the axioms that define what it means to have an addition and a multiplication that cooperate.

This is why the rule shows up unchanged in places where the words “negative” and “positive” have no ordinary meaning at all.

Ring	Example of the rule	Check
Integers $\mathbb{Z}$	$(-3)(-4) = 12$	Familiar
Clock arithmetic $\mathbb{Z}/7$	$(-2)(-3) = 5\cdot 4 = 20 \equiv 6$	And $2\cdot 3 = 6$
Polynomials $\mathbb{R}[x]$	$(-x)(-x+1) = x^2 - x$	Expand and compare
$2\times 2$ matrices	$(-A)(-B) = AB$	Even though $AB \neq BA$ in general
Continuous functions	$(-f)(-g) = fg$ pointwise	Pointwise product

The clock arithmetic row is worth pausing on. In $\mathbb{Z}/7$ there is no order, no left and right, no notion of a number being “less than nothing”. The element $-2$ simply means the element that adds to $2$ to give $0$, which is $5$. And indeed $5 \cdot 4 = 20 = 2\cdot 7 + 6 \equiv 6$, which is exactly $2 \cdot 3$. The sign rule survives in a setting where every intuitive picture has been stripped away, which is fairly strong evidence that the pictures were never what was holding it up.

The matrix row makes a related point. Matrix multiplication is not commutative, so a great deal of ordinary algebraic intuition fails for matrices. The sign rule does not fail, because its proof never used commutativity of multiplication.

Historically this insight was slow in coming. George Peacock, in his Treatise on Algebra of 1830 and in a report to the British Association a few years later, proposed what he called the principle of the permanence of equivalent forms: an identity established for the restricted, intuitive number system should be carried over as a definition when the system is enlarged. Hermann Hankel sharpened this in 1867, restricting the permanence claim to the formal laws themselves rather than to every theorem. On that view, the sign rule is exactly what the permanence of the distributive law compels. Modern axiomatic algebra is the mature form of the same idea.

The models that actually work

With the mathematics settled, the pictures can be enjoyed for what they are.

Removing debts. Suppose a subscription charges you 20 euros per month and the charge appears in your ledger as $-20$. Five months of it is $5 \times (-20) = -100$. Now suppose the company discovers a billing error and reverses five of those charges. Removing five charges is $-5$ of them, so the effect on your balance is

$(-5)\times(-20) = +100.$

Your balance goes up by 100 euros. The two negatives are genuinely independent: one is the sign of the charge, the other is the direction of the ledger operation. This example works, but note how much of the work is done by the word removing, which is where the second reversal is hidden.

Velocity and time. Take a position axis with positive to the east, and let a cyclist travel at $v = -3$ metres per second, meaning three metres per second westward. Position is $x(t) = x_0 + vt$, with $t$ measured from now. Where was the cyclist four seconds ago? Put $t = -4$:

$x(-4) - x_0 = (-3)\times(-4) = +12.$

Twelve metres to the east. This is not a trick. It is the correct physical answer, and you can verify it by watching: someone heading west is, at earlier times, further east. Reversing the direction of motion and reversing the direction of time both flip the sign, and doing both returns you to where you started.

Reflection on the number line. This is the cleanest of the three. Multiplication by $-1$ is the map $x \mapsto -x$, which geometrically is reflection of the line through the point $0$. Apply it once and $3$ goes to $-3$. Apply it again and $-3$ goes back to $3$. A reflection composed with itself is the identity, which is exactly the statement $(-1)\cdot(-1) = 1$.

This is the version that scales. Multiplication by a positive number $k$ stretches the line by a factor of $k$. Multiplication by $-k$ stretches by $k$ and then reflects. Composing two such operations, $(-a)\cdot(-b)$, stretches by $ab$ and reflects twice, and two reflections leave the line as it was. The sign rule is the statement that the reflections cancel.

Rotation, and the picture that also explains $i$

The reflection picture has a sequel. On the real line, reflection through the origin is the only motion available that reverses direction. In the plane, the same map $x \mapsto -x$ can be seen differently: it is rotation by $180^\circ$ about the origin.

That reframing is more powerful, because rotations compose by adding angles. Multiplying by $-1$ rotates by $180^\circ$; multiplying by $-1$ twice rotates by $360^\circ$, which is the identity. So

$(-1)\cdot(-1) = 1$

becomes the statement that half a turn followed by half a turn is a full turn.

Once you have that, an obvious question presents itself: what number rotates by $90^\circ$? Whatever it is, doing it twice must give a $180^\circ$ rotation, so its square must be $-1$. That number is $i$, and the entire apparatus of complex numbers unfolds from the observation. Multiplication by $i$ is a quarter turn: $1 \mapsto i \mapsto -1 \mapsto -i \mapsto 1$, and $i^4 = 1$ is the statement that four quarter turns bring you home.

There is a satisfying moral here. The rule that students find most arbitrary, $(-1)(-1) = 1$, is the degenerate case of the most geometrically natural fact in all of algebra: multiplying complex numbers adds their angles. The negatives are simply the numbers whose angle is $180^\circ$, and $180 + 180 = 360$.

The history: from absurd to obligatory

Negative numbers were resisted for a very long time, and it is worth being precise about who resisted what.

Diophantus of Alexandria, writing in the third century, worked only with positive rational solutions. In the Arithmetica he encounters the equation that we would write as $4x + 20 = 4$ and dismisses it as absurd, because its only solution is $-4$. He had a symbol for subtraction and used it freely; what he did not have was the idea that a negative quantity could stand alone as an answer.

The first systematic treatment came from India. In the Brahmasphutasiddhanta of 628, Brahmagupta laid out arithmetic rules for what he called fortunes (positive) and debts (negative), together with zero. Among them is precisely the statement at issue: the product of two debts is a fortune. Brahmagupta gave the rule; he did not derive it from more primitive principles, and there was no reason at the time to think a derivation was needed.

Europe took much longer. Cardano in the sixteenth century called negative solutions fictitious. Descartes, in La Géométrie of 1637, called them false roots, and his original coordinate diagram was a single quadrant. Simon Stevin, working around 1594, still separated the quadratic equation into three distinct types so as to keep all coefficients positive, following al-Khwarizmi’s classification of seven centuries earlier.

Colin Maclaurin’s Treatise of Algebra, published posthumously in 1748, argues that negative quantities are no less real than positive ones. That he had to argue the point tells you it was still live. Maria Gaetana Agnesi’s textbook of the same year, 1748, takes the modern view directly, treating negative roots as having the same status as positive ones and differing only in the direction measured from the origin.

The hostility did not die quietly. In 1796 William Frend published The Principles of Algebra, in which he denied outright that negative numbers exist, holding that whenever such an expression appeared, the error lay either with the person who proposed the problem or with the person attempting to solve it. Frend was not a crank on the margins; his daughter Sophia married Augustus De Morgan in 1837. And De Morgan himself, as we saw, still described $3-8$ as absurd in 1831, while simultaneously constructing a careful framework for using such expressions safely: a negative answer, he wrote, indicates that the nature of the answer is the reverse of what was supposed.

What finally resolved matters was not a better intuition but a change of question. Instead of asking what a negative number is, nineteenth-century algebraists asked what laws the symbols should obey, and defined the objects to be whatever satisfies those laws. Once the question is posed that way, the answer to our title question is immediate: $(-1)(-1) = 1$ is not a claim about reality that might have gone another way. It is a consequence of demanding that algebra remain algebra.

What to tell a student who is not satisfied

The unsatisfying answers are unsatisfying for a reason that can be stated honestly, and stating it honestly usually helps more than another analogy.

Negative numbers do not describe quantities of things. They describe positions relative to a reference point and operations that reverse direction. There is no arrangement of physical objects that exhibits a negative times a negative, and hunting for one is a category error, not a failure of imagination.

What there is instead is a demand for consistency. Arithmetic with positive numbers obeys a small set of laws, the most important being that multiplication distributes over addition. When the number system is extended to include negatives, one could in principle assign any value at all to $(-1)\cdot(-1)$. But exactly one assignment preserves the laws, and every other choice destroys the ability to expand a bracket. Mathematics chose consistency, and the sign rule is what consistency cost, which is to say it cost nothing at all.

The rule is then vindicated by the company it keeps: it is the same rule that makes reflections cancel, that makes $180^\circ$ plus $180^\circ$ a full turn, that makes $i^2 = -1$ sensible, and that holds verbatim in matrix rings and clock arithmetic where none of the everyday pictures apply. A rule that arbitrary would not survive that many contexts unchanged.

---

Related topics: Imaginary Numbers: Why the Square Root of -1 Turned Out to Matter, Modular Arithmetic: The Mathematics of Clocks and Cryptography, What Matrices Really Are, The Beauty of Symmetry: A Short Tour of Group Theory, Vector Spaces: The Architecture of Mathematics, Quaternions: The 4D Numbers That Power 3D Graphics, Surreal Numbers: The Number System Born in a Game of Go, The Fundamental Theorem of Algebra: Why Every Polynomial Has a Root

Frequently asked

Why does minus times minus equal plus?

Because the distributive law forces it. Start from the fact that any number times zero is zero, so (-1) times (1 + (-1)) equals 0. Distributing gives (-1)(1) + (-1)(-1) = 0, which is -1 plus (-1)(-1) = 0. The only value of (-1)(-1) that makes this true is 1. If you chose -1 instead, the distributive law would fail somewhere else in arithmetic.

Is there a real world example of negative times negative?

Yes, though none of them are as immediate as three baskets of four apples. Cancelling debts is the standard one: removing five charges of 20 each raises your balance by 100. Velocity and time also work: an object moving at -3 metres per second was, 4 seconds ago, 12 metres in the positive direction. The cleanest picture is geometric, since multiplying by -1 reflects the number line through zero, and reflecting twice puts everything back.

How do you prove that minus 1 times minus 1 equals 1?

In three lines. First, a times 0 equals 0 for every a, because a times 0 equals a times (0 + 0) equals a times 0 plus a times 0. Second, apply this with a equal to -1 and write 0 as 1 + (-1). Third, distribute and use the fact that -1 times 1 is -1. You are left with -1 plus (-1)(-1) equals 0, so (-1)(-1) equals 1.

Why is there no good real life example of this?

Because multiplication of two quantities that are both counts of physical objects can never involve two negatives. Negative numbers describe direction, debt, deficit or displacement, not amounts of stuff. Any honest example therefore has to involve two independent reversals, and a situation with two reversals is harder to picture than a situation with none. That is a limitation of intuition, not of the mathematics.

Is the negative times negative rule just something someone made up?

No. It is the only choice consistent with the laws of addition and multiplication that already hold for positive numbers. Once you insist that the distributive law survives the extension to negative numbers, the sign rule is a theorem, not a convention. It holds in every algebraic structure that satisfies the ring axioms, including matrices, polynomials and clock arithmetic.

Continue reading

Blog

Vector Spaces: The Architecture of Mathematics

Polynomials, functions, signals, and forces all live in the same kind of mathematical structure: a vector space. Understanding this abstract framework unifies enormous swaths of mathematics.

Read →Blog

What Are the Roots of Unity? Geometry, Cyclotomy and Fourier

What are the roots of unity, why there are exactly phi(n) primitive ones, how cyclotomic polynomials factor x^n-1, and why they make the FFT and the 17-gon work.

Read →Blog

Why Do People Get Different Answers to 8÷2(2+2)?

People get different answers to 8÷2(2+2) because the expression is ambiguously written, not because anyone has forgotten arithmetic. Here is what is really going on.

Read →