Can We Build All Math from Pure Thinking Alone?
A Letter That Changed Everything
In June 1902, the German mathematician Gottlob Frege (1848–1925) was finishing the second volume of his masterwork, the Grundgesetze der Arithmetik—the “Basic Laws of Arithmetic.” He believed he had finally shown that all of math could be reduced to pure logic. Then a letter arrived from a young British thinker, Bertrand Russell (1872–1970). Russell had found a hidden contradiction in one of Frege’s cornerstone rules, Basic Law V. That single crack was enough to collapse the whole system.
Frege was devastated. He added a rushed appendix admitting the problem. But even in that wreckage lay a startling fact: Frege had already proved that the fundamental rules of numbers—the ones we use every day—could be derived from a different, perfectly consistent principle. This proof, now called Frege’s Theorem, didn’t need the broken law at all. It raised a question that still bothers philosophers today: what are numbers, and how do we know about them?
The Dream: Turning Math into Logic
Frege’s goal was logicism—the idea that arithmetic is just a branch of logic. If you could start with a handful of logical truths and definitions, you could build up 2+2=4 and every other number fact without appealing to experience or intuition. It would mean math is as certain as “not both P and not P.”
To do this, Frege invented a whole new language. In ordinary grammar, we say “Socrates is mortal.” Frege broke this into a logical skeleton: there’s an object (Socrates) and a concept (the property “being mortal”). A concept is like a mathematical function that, when you feed it an object, spits out a truth value—The True or The False. Feed it Socrates, and “is mortal” returns The True. Feed it a rock, The False. In Frege’s language, you could write a formula like M(s) to mean “object s falls under concept M.”
He also allowed second‑order logic, which lets you talk about concepts the same way you talk about objects. You can say “Everything that is a human is mortal” with quantifiers over objects, but also “There is a property that all numbers share” with quantifiers over concepts. And with a rule called the Comprehension Principle for Concepts, you can always form a concept from any logical condition. For example, from “x is round and x is red” you can name the concept round‑and‑red. Frege even used a device like modern lambda (λ) notation to write this compactly: [λx (Round(x) & Red(x))] names that complex concept.
So far, so good. The language was powerful but seemed safe. Now Frege needed one more piece to get from logic to numbers: a way to turn concepts into objects he called extensions.
The Dangerous Law
An extension of a concept is like the set of all objects that fall under it. The extension of “is a positive even integer less than 8” would be the collection {2, 4, 6}. Frege introduced an operator ε to name extensions. For any concept F, εF is an object—the extension. And he needed a law to govern when two extensions are identical.
Basic Law V stated: εF = εG if and only if exactly the same objects fall under F and G (i.e., ∀x(Fx ↔ Gx)). In plain English, two extensions are equal just when the concepts are materially equivalent. The idea was that concepts and extensions would pair up in a disciplined way.
But Russell’s letter showed this law leads straight to paradox. Using Basic Law V plus the Comprehension Principle, you could form the concept “is an extension of a concept that it does not fall under.” Call that concept R. Ask: does εR fall under R? If yes, then by definition it doesn’t; if it doesn’t, then it does. It’s like the barber who shaves all and only those who don’t shave themselves—an impossible situation.
Why did it break? Basic Law V demands that the domain of concepts be no larger than the domain of extensions. But the Comprehension Principle keeps producing new concepts for every condition, forcing concepts to outnumber extensions. The two demands can’t both be satisfied. The law was simply too strong.
The Strange Power of Counting Equally
Frege had already sensed a safer path. Years earlier he had used a different principle, now called Hume’s Principle (after the philosopher David Hume, though Frege formalized it). It says: the number of Fs equals the number of Gs if and only if the Fs and Gs are equinumerous—that is, can be matched one‑to‑one.
To make this precise, Frege defined equinumerosity using relations. Two concepts F and G are equinumerous if there exists a relation R that pairs every object falling under F with exactly one object falling under G, and vice versa, like a perfectly paired dance where no one is left without a partner. Using “#F” for “the number of Fs,” Hume’s Principle becomes: #F = #G ↔ F ≈ G.
Unlike Basic Law V, Hume’s Principle doesn’t demand a one‑to‑one matchup between concepts and numbers—only between concepts that are already equinumerous. Many different concepts can have the same number; the concept “author of Principia Mathematica” (Russell and Whitehead) and the concept “positive integer between 1 and 4” (2 and 3) both have the number 2. Hume’s Principle is consistent with second‑order logic. That makes it a safe place to stand.
Building Numbers from Scratch
Here is the amazing part: Frege didn’t need extensions at all to build the familiar natural numbers. With Hume’s Principle and second‑order logic, he defined each piece.
He started with zero: 0 = #F, where F is the concept “being non‑self‑identical” (something nothing can be). Then he defined the predecessor relation: one number immediately comes before another. For example, 1 precedes 2 if there’s a concept F and an object a such that a falls under F, the number of Fs is 2, and the number of “F other than a” is 1. This works perfectly.
Next, he needed a way to chain steps. He invented the ancestral of a relation—a concept we use all the time. In a family tree, “is a parent of” is a relation; its ancestral “is an ancestor of” links you to grandparents and beyond, following the relation any number of times. Frege used this to define the natural numbers as anything you can reach from 0 by following the predecessor relation zero or more times.
From this foundation, he proved the classic Peano axioms: 0 is a natural number; 0 is not the successor of any natural number; no two numbers share a successor; mathematical induction works; and every natural number has a successor. In other words, all ordinary arithmetic—addition, multiplication—grows out of these logical roots. The proof is rigorous. Frege had, without the flawed Basic Law V, derived the laws of numbers.
What Is a Number, Anyway?
So Frege succeeded in a way, but a huge philosophical puzzle remains: what sort of thing is a number? Hume’s Principle tells you when two numbers are equal, but it never says whether Julius Caesar is the number 2. Could the Roman dictator literally be a number? Obviously not, but the principle alone doesn’t rule it out. This is the Julius Caesar problem, named after Frege’s own joke.
It points to a deeper worry: how do we even know that numbers exist? Hume’s Principle implies that for any concept, there is a number—an object. But is that an analytic truth of logic, or do we need some special mental “intuition” to see it? Philosophers still argue. Frege himself thought his explicit definition of numbers as extensions would solve everything; that path collapsed. Modern logicians often separate existence claims from identity conditions, but the debate over what grounds our knowledge of abstract objects is very much alive.
Frege’s Theorem shows that arithmetic can be anchored in a consistent logical principle. It doesn’t answer the deeper “what are numbers” question, but it hands us a precise map of what must be explained. Every time you count your change or add 7 and 5, you’re relying on laws whose logical skeleton Frege first laid bare.
Think about it
- If numbers are just ideas in our heads, why does 2+2 always equal 4 for everyone, everywhere?
- Could there be a world where 2+2 wasn’t 4, and if so, what would that say about what numbers really are?
- Suppose you can pair every spoon in a drawer with a fork without leftovers, even if you never count them. How could you be sure the number of spoons equals the number of forks? What does that tell you about what “number” means?
