We measure how belief is made: in a story, in an idea, in a live event.

← All essays
Cosmology

When the Universe Runs Out Before the Numbers Do

From counting on our fingers to the point where number names become a burden. How do we understand numbers larger than the stars and atoms we can count?

2026-11-16 · A.F. Sadek

Prefer reading in Arabic? Read the Arabic version.

“How high can you count?”

A child starts eagerly: one, ten, a hundred, a thousand, a million… Then comes the biggest word they know, and they assume the story ends there. As adults, we reach for heavier names: billion, trillion, quadrillion. Then we drop “infinity” into the sentence to close the subject.

But knowing a name is a long way from feeling its size. Each zero added on the right multiplies the number by ten; three zeros multiply it by a thousand. As we climb from 1, long names become less useful, comparisons with things around us become harder, and we turn to rules that describe numbers instead of writing out every digit. Our way of representing the number changes. The counting line stays open.

(Scale note: this article uses the modern short scale: million = 10⁶, billion = 10⁹, trillion = 10¹². In the Arabic version, billion is milyar; the term bilyun is avoided because Arabic usage can assign it either nine or twelve zeros. Beyond trillion, the Arabic text pairs English names with powers of ten.)

The seconds ruler: beyond the age of the universe

To feel what three more zeros do, let us count a familiar unit: one second.

A million seconds (10⁶) is about 11.57 days, a little less than two weeks.

Multiply by a thousand and we reach a billion seconds (10⁹): those two weeks stretch into 31.69 years.

Multiply by a thousand again for a trillion seconds (10¹²): about 31,689 years, deep into prehistory, before farming and cities.

The next step, a quadrillion seconds (quadrillion = 10¹⁵), gives 31.69 million years. Then comes the surprise: a quintillion seconds (quintillion = 10¹⁸) is 31.69 billion years.

Stop there. Standard cosmological estimates put the age of the universe at about 13.8 billion years.1

A counter that has barely reached eighteen zeros, 10¹⁸ seconds, the second named step after a trillion, has already passed twice the age of the universe. To continue our comparison, let us move from counting time to counting matter.

Counting matter: stars, your body, and Earth

Broad estimates put the number of stars in the observable universe between 10²² and 10²⁴, from ten sextillion to one septillion.2 The upper estimate is one hundred times the lower: the exponents differ by two, and 10² = 100.

Look inside our bodies instead. A person weighing 70 kilograms is estimated to contain about 7 × 10²⁷ atoms, or 7 octillion.3 Your body holds thousands of times more atoms than there are stars in the observable sky.

Now expand the view to Earth. A calculation based on its mass and composition gives about 1.4 × 10⁵⁰ atoms: an order of magnitude of 10⁵⁰, below sexdecillion, a one followed by 51 zeros.4

The Latin numbering behind the names

  • Million: first in this ladder; the word itself has a separate origin.
  • billion: “bi” means two.
  • trillion: “tri” means three.
  • vigintillion: “viginti” means twenty, so this is name number 20.

To find the number of zeros from the name’s position, multiply by 3 and add 3. For vigintillion (number 20), 3 × 20 + 3 gives 63 zeros. For centillion (number 100), 3 × 100 + 3 gives 303 zeros.

We do not ask for a vigintillion of anything in everyday life. Yet the -illion system can continue, and names can be constructed for much higher powers.

At 10⁸⁰: the limits of comparison with matter

Gather all the ordinary atoms in the observable universe and we reach the familiar order of magnitude of 10⁸⁰ atoms. That lies between quinvigintillion (10⁷⁸) and sesvigintillion (10⁸¹).5

Light takes the count higher. The cosmic microwave background (CMB) contains about 1.5 × 10⁸⁹ photons within the observable universe. Photons greatly outnumber atoms, but the count still falls between octovigintillion (10⁸⁷) and novemvigintillion (10⁹⁰).6

At these scales, powers of ten are easier to use than a string of long names. Those names become less useful well before we reach such heights.7

The naming system can keep going. Conway and Guy set out a systematic way to construct names beyond even centillion (10³⁰³). Language can produce more names; their practical value declines as they grow longer.8

Googol: a guest inside the ladder

Some imagine a googol (googol = 10¹⁰⁰) at the end of all the names, on the edge of eternity. It is a 1 followed by a hundred zeros, and its place is inside the ladder, between duotrigintillion (10⁹⁹) and trestrigintillion (10¹⁰²).

The word first appeared in print in a paper by Kasner in Scripta Mathematica (1938), then became widely known through Kasner and Newman’s book (1940). It was proposed by Kasner’s young relative Milton Sirotta, aged nine. Milton’s brother Edwin Sirotta’s family has also credited both brothers with inventing the name.9

The comparison with physics needs clear boundaries:

  • A googol exceeds the number of atoms (~10⁸⁰) and CMB photons (~1.5 × 10⁸⁹).
  • It does not exceed every physical count. Divide the observable universe’s volume by the cube of the Planck length and the result is about 8 × 10¹⁸⁴ (~10¹⁸⁵) Planck volumes.10

The resulting number is a ratio between two volumes. A Planck volume is the cube of the Planck length. Using it does not mean that space consists of separate cells of that size.

Googolplex: where paper and atoms run out

A hundred zeros can fit on a couple of lines. Kasner’s circle introduced the googolplex (googolplex = 10^googol = 10^(10¹⁰⁰)): a one followed by a googol of zeros.

Kasner and Newman’s famous story shows why a number cannot be defined by how long someone can keep writing zeros. Our own comparison makes the scale concrete: if we wrote one digit on each atom in the observable universe (~10⁸⁰), we would run out of atoms before writing more than a tiny fraction of a googolplex’s zeros.11

Later came googolplexian (10^googolplex), an informal name documented in literary and everyday use. It was not coined by Kasner and is not a major-dictionary standard. Even this huge tower of powers is small beside what rules can produce.12

Beyond writing out digits: when a recipe makes the number

Now we leave strings of zeros and enter the world of recipes, with an important warning: Graham’s number, TREE(3), and Rayo’s number are not consecutive stops just before infinity. Each comes from a different problem. We arrange them here only to compare their size (Graham < TREE(3) < Rayo). All are strictly finite.

  • Graham’s number (G): popularized by Gardner (1977), this is distinct from the bound N in Graham and Rothschild’s paper (1971) on coloring pairs of vertices of a high-dimensional hypercube with two colors. The observable universe could not hold all of Graham’s number’s digits even if each digit occupied one Planck volume. Yet a finite recursive rule defines it.13
  • TREE(3): described by Friedman (2006) on the Foundations of Mathematics mailing list through a three-colour tree game. Tree i has at most i vertices, and no earlier tree may embed in a later one while preserving colors and branching. The theorem guarantees that the game must end after finitely many trees. Its maximum length, TREE(3), nevertheless dwarfs Graham’s number and remains finite.14
  • Rayo’s number (Rayo(10¹⁰⁰)): introduced on 26 January 2007 in MIT’s “Big Number Duel” between philosophers Agustín Rayo and Adam Elga. It is the smallest finite number greater than every finite number that can be uniquely named in the language of first-order set theory with fewer than a googol (10¹⁰⁰) symbols. Its formulation still raises precise philosophical questions.15

Why N + 1 always wins, and where infinity belongs

Take any finite number we have reached: centillion, googolplex, G, TREE(3), or even Rayo’s number.

Call it N.

What stops you from writing N + 1?

Nothing. Every positive whole number has another number one greater than it. The counting sequence therefore has no final number.

Here is the distinction: infinity is not a number that comes after the last finite number. However many times we add one, a finite number of steps gives another finite number. There is no final step that takes us from counting into infinity.16

In the film, the song jumps between the major families of names. The caption names the song skips makes those omissions explicit, preserving the pace without suggesting that every name has been sung. Watch the film.

We began with fingers, then borrowed seconds, stars, matter, and light to give the numbers scales we could picture. We outgrew every benchmark we chose and reached numbers defined by rules because their digits would overwhelm any writing medium we imagined. Yet one move remains available, however large the number: add one more.

In our research sample of 21 videos, 6 placed infinity directly after finite numbers. We also read corrections published by Reigarw and Numberphile concerning mathematical information in their videos. In Arabic material, we found an article that mixes the two meanings of bilyun and describes a googolplex as “the largest number of all.”17

Notes and references

  1. Conversions from seconds to days and years are approximate. The cosmic age comes from PDG 2024, Astrophysical Constants and Parameters, revised August 2023, reporting Planck 2018 values: 13.797 ± 0.023 billion years, or about 13.787 ± 0.020 with BAO data. The main text uses the rounded value of 13.8 billion years.

    ↩
  2. European Space Agency: How many stars are there in the Universe?. The 10²²–10²⁴ range is a broad estimate, not a direct census of every star.

    ↩
  3. Jefferson Lab: How many atoms are in the human body?, assuming a 70 kg body; and Freitas, Nanomedicine, volume I (1999), §3.1. The estimate varies with body mass and composition.

    ↩
  4. The calculation uses Earth’s mass of 5.972 × 10²⁴ kg from PDG 2024, McDonough’s (2003) bulk-Earth composition model, and the Avogadro constant of 6.02214076 × 10²³ mol⁻¹, exact by definition in the SI since 2019 (BIPM), giving about 1.4 × 10⁵⁰. An educational presentation held by NASA NTRS, record 20210010963, slide 65, gives 3 × 10⁵⁰: it divides all of Earth’s mass by carbon’s molar mass (12 g/mol) as a quick approximation. The estimates differ by roughly a factor of two on a ladder whose rungs differ by a thousand.

    ↩
  5. NASA SP-4802, Cosmos & Culture uses the 10⁸⁰ scale in the context of counting protons and neutrons. That differs from a census of neutral atoms: most ordinary matter is ionized, and a helium nucleus has four nucleons. An independent calculation using Ω_b ≈ 0.049, H₀ ≈ 67.7 km/s/Mpc, a hydrogen mass fraction of 0.75, and a radius of 46.5 billion light-years gives an equivalent atomic-nucleus count of about 7.3 × 10⁷⁹. We use 10⁸⁰ as an approximate order of magnitude, not an exact count.

    ↩
  6. This calculation concerns CMB photons, using Fixsen’s (2009) temperature of 2.72548 ± 0.00057 K, PDG 2024’s density of about 411 cm⁻³, and the observable universe’s volume. The 46.1–46.6 billion light-year radius range is supported by Gott et al. (2005), Lehners and Quintin (2024), and Planck 2018 parameters. The result is about 1.5 × 10⁸⁹ photons.

    ↩
  7. NIST SP 811 (2008), §7.10.3 recommends avoiding number names from 10⁹ upward in scientific expressions because their meanings vary between scales. The familiar dictionary range, million through vigintillion and then centillion, describes the selection of entries, not a mathematical limit; see the comparison of names and dictionary coverage.

    ↩
  8. John H. Conway and Richard K. Guy, The Book of Numbers, 1996. Examples of constructed names include tresvigintillion, sesvigintillion, septemvigintillion, trigintillion, duotrigintillion, and trestrigintillion.

    ↩
  9. David Wilton’s account in Wordorigins traces the word’s print appearance to Edward Kasner, “New Names in Mathematics,” Scripta Mathematica 5(1) (1938), p. 13; then Kasner and James Newman, Mathematics and the Imagination (1940). The account of Edwin’s participation comes from Carl Bialik, “There Could Be No Google Without Edward Kasner,” The Wall Street Journal (2004).

    ↩
  10. The Planck length is 1.616 255(18) × 10⁻³⁵ m, a CODATA 2022 value also listed in PDG 2024. Dividing the observable universe’s volume by the cube of this length gives about 8 × 10¹⁸⁴, an order of magnitude of 10¹⁸⁵. The comparison does not assume that these volumes are actual cells of space.

    ↩
  11. Kasner and Newman’s story contrasts Einstein’s and Carnera’s endurance in writing zeros to show why a number’s definition cannot depend on when its writer gets tired. The main text’s physical argument assumes one digit per atom: the 10¹⁰⁰ required places exceed ~10⁸⁰, but fall below the Planck-volume count calculated above. That does not establish that a real digit could be stored in a Planck volume.

    ↩
  12. Wiktionary: googolplexian, including literary and informal examples from 1989 and 2013. We present it as an informal name for 10^googolplex, not a standard name coined by Kasner.

    ↩
  13. Martin Gardner, Scientific American 237(5) (1977), pp. 18–28; and R. L. Graham and B. L. Rothschild, “Ramsey’s Theorem for n-Parameter Sets” (1971). The main text refers to G, popularized by Gardner, rather than the paper’s bound N. For the comparison with the space needed to write it out, see MathWorld’s account of Graham’s number.

    ↩
  14. Harvey Friedman, “273:Sigma01/optimal/size,” FOM, 28 March 2006. The definition requires embeddings that preserve labels and the points where branches meet. TREE measures the longest allowed sequence, not the size of one tree.

    ↩
  15. Agustín Rayo’s original Big Number Duel page. The formal definition uses fewer than a googol symbols in first-order set theory, with a semantic interpretation formulated using second-order tools. The short account in the main text retains those language and interpretation conditions. Increasing the symbol budget by one does not by itself guarantee a strictly larger number. The contestants agreed to avoid winning simply by adding one to an opponent’s number; a new entry had to go beyond what could practically be reached by earlier methods. That restriction belongs to the contest, not to mathematics: N + 1 remains greater than N for any natural number N.

    ↩
  16. Mathematics uses cardinal numbers to measure the sizes of infinite sets and ordinal numbers to describe their order types. These concepts differ from reaching infinity by adding one to a finite number.

    ↩
  17. The project’s survey and its review are dated 14 October 2026. The figures describe that sample, not all available content. The research review checked Reigarw’s correction list and the description of Numberphile’s “How Big is Graham’s Number.” The Arabic example is Ra2ej’s “Names of numbers after a million”, which attributes the name to Kasner’s infant son and mixes two meanings of bilyun. These are different errors; we do not attribute all the channels’ corrections to the infinity misconception.

    ↩