Some questions in mathematics look different on the surface but collapse to the same proof underneath. "Is 0.999\ldots = 1?" feels philosophical. "Does 0.333\ldots = 1/3?" feels like a calculator check. "Is 0.142857\ldots = 1/7?" feels like a pattern-spotting puzzle. They are all the same question wearing three different costumes.

The moment a problem asks whether a repeating decimal equals a simple closed-form number — a fraction, an integer, a ratio — your brain should fire a single pattern:

Repeating decimal = simple number? → prove it by (a) supremum, or (b) geometric series.

That is the entire playbook. Two routes, both rigorous, both short. This page is the drill that trains the recognition and the two templates.

The pattern

Any repeating decimal 0.d_1 d_2 d_3 \ldots is, by definition, a limit of finite truncations:

0.d_1 d_2 \ldots d_n \to 0.d_1 d_2 d_3 \ldots \quad \text{as } n \to \infty.

You cannot "evaluate" an infinite string of digits directly. What the notation 0.999\ldots actually means is the real number that the sequence 0.9, 0.99, 0.999, \ldots converges to. The question "does this equal 1?" is the question "is 1 the limit?" And there are only two clean ways to pin down that limit:

(a) Supremum via completeness. The truncations form an increasing sequence bounded above. By the least upper bound property, the supremum exists as a real number. Show it equals your target.

(b) Geometric series. The difference between consecutive truncations forms a geometric progression. The sum of the infinite progression is given by the closed formula \frac{a}{1 - r}. Evaluate it; compare to your target.

Every "is 0.\overline{abc\ldots} = p/q?" question yields to one of these. Sometimes both.

Route (a): the supremum argument

The cleanest proof of 0.999\ldots = 1 in one line of reasoning goes like this.

Let S = \{0.9,\ 0.99,\ 0.999,\ 0.9999,\ \ldots\}. This is a non-empty subset of \mathbb{R}. Every element is less than 1, so 1 is an upper bound. By completeness, \sup S exists. Call it L.

You want to show L = 1. Two steps.

L \le 1. Since 1 is an upper bound and L is the least upper bound, L \le 1.

L \ge 1. Suppose L < 1. Then 1 - L > 0. By the Archimedean property, pick n with 10^n > \frac{1}{1 - L}, i.e. \frac{1}{10^n} < 1 - L. Then 1 - \frac{1}{10^n} > L. But 1 - \frac{1}{10^n} is the n-th element of S (it equals 0.\underbrace{99\ldots9}_{n}). So S contains an element strictly greater than L — contradicting L being an upper bound.

So L = 1, and the notation 0.999\ldots, which is \sup S by definition, equals 1.

That argument is the skeleton for every "repeating decimal = ?" problem. You identify the truncation sequence, establish it is bounded, invoke completeness, and pin the supremum to the target value via the Archimedean squeeze.

Route (b): the geometric series argument

The geometric series route is more computational and often faster when the repeating block is simple.

Write the repeating decimal as an infinite sum:

0.999\ldots = \frac{9}{10} + \frac{9}{100} + \frac{9}{1000} + \cdots = \sum_{n=1}^{\infty} \frac{9}{10^n}.

This is a geometric series with first term a = \frac{9}{10} and common ratio r = \frac{1}{10}. Since |r| < 1, the series converges, and the sum is

\frac{a}{1 - r} = \frac{9/10}{1 - 1/10} = \frac{9/10}{9/10} = 1.

Done. No \varepsilon, no supremum. Just the closed-form formula applied once.

The catch: "the series converges" is itself a completeness statement. The formula \frac{a}{1-r} is a theorem that holds because \mathbb{R} is complete. Route (b) looks easier because the completeness is hidden inside the formula — but it is still doing the work.

Side-by-side: which to reach for

Side-by-side comparison of the supremum route and the geometric series route for proving 0.999 repeating equals 1Two stacked boxes separated by a vertical divider. The left box is titled 'Route (a): Supremum' and lists four steps: list truncations as a set S, note S is bounded above by 1, invoke completeness so sup S exists, use the Archimedean property to show sup S equals 1. The right box is titled 'Route (b): Geometric series' and lists four steps: write the decimal as an infinite sum of terms 9 over 10 to the n, identify first term a equals 9 over 10 and ratio r equals 1 over 10, apply the sum formula a over 1 minus r, simplify to 1. At the bottom a note reads: both routes use completeness; route b hides it inside the sum formula. Route (a): Supremum Route (b): Geometric series 1. Let S = {0.9, 0.99, 0.999, …} 2. S bounded above by 1 3. Completeness ⇒ sup S exists 4. Archimedean squeeze ⇒ sup S = 1 Feels: foundational, axiomatic Uses: LUB property directly Best when: you need to defend the equality from scratch 1. Write as Σ 9/10ⁿ for n ≥ 1 2. a = 9/10, r = 1/10, |r| < 1 3. Apply sum = a / (1 − r) 4. (9/10) / (9/10) = 1 Feels: computational, quick Uses: geometric sum formula Best when: you just want the answer fast and cleanly Both rest on completeness — route (b) hides it inside the sum formula.
The two routes are fully equivalent in logical strength. Route (a) is the right one to reach for in a proof-based course or when a skeptic presses you on why the equality holds. Route (b) is the one to deploy when you want the arithmetic done in three lines.

Worked example 1: 0.333\ldots = \tfrac{1}{3}

Route (a): supremum. Let S = \{0.3,\ 0.33,\ 0.333,\ \ldots\}. Each element is \frac{1}{3} - \frac{1}{3 \cdot 10^n} — verify: 0.3 = \frac{3}{10} = \frac{1}{3} - \frac{1}{30}. All elements are less than \frac{1}{3}, so \frac{1}{3} is an upper bound, so \sup S \le \frac{1}{3}. Conversely, for any L < \frac{1}{3}, the Archimedean property gives n with \frac{1}{3 \cdot 10^n} < \frac{1}{3} - L, i.e. \frac{1}{3} - \frac{1}{3 \cdot 10^n} > L, so the n-th element of S exceeds L. Hence \sup S = \frac{1}{3}.

Route (b): geometric series.

0.333\ldots = \frac{3}{10} + \frac{3}{100} + \frac{3}{1000} + \cdots = \frac{3/10}{1 - 1/10} = \frac{3/10}{9/10} = \frac{3}{9} = \frac{1}{3}.

Two lines. Done.

Worked example 2: 0.\overline{142857} = \tfrac{1}{7}

The repeating block has length 6. Route (b) handles this cleanly; route (a) is tedious for long blocks, so this is the right place to favour the series.

Route (b): geometric series. Group the digits in blocks of six:

0.\overline{142857} = \frac{142857}{10^6} + \frac{142857}{10^{12}} + \frac{142857}{10^{18}} + \cdots

First term a = \frac{142857}{10^6}, common ratio r = \frac{1}{10^6}. The sum is

\frac{a}{1 - r} = \frac{142857 / 10^6}{1 - 1/10^6} = \frac{142857 / 10^6}{(10^6 - 1)/10^6} = \frac{142857}{10^6 - 1} = \frac{142857}{999999}.

Now simplify: 999999 = 7 \times 142857 (a well-known identity — try it on a calculator). So \frac{142857}{999999} = \frac{142857}{7 \times 142857} = \frac{1}{7}.

Route (a): supremum. Let S_n be the truncation 0.\underbrace{142857\,142857\,\ldots\,142857}_{n \text{ blocks}}. Then S_n = \frac{1}{7}(1 - 10^{-6n}) (verify by multiplying \frac{1}{7} by 1 - 10^{-6n} and noting the repeating expansion). The set S = \{S_n\} is bounded above by \frac{1}{7}, and for any L < \frac{1}{7}, the Archimedean property gives n with \frac{1}{7} \cdot 10^{-6n} < \frac{1}{7} - L, making S_n > L. So \sup S = \frac{1}{7}.

Both routes land. The geometric series is three lines; the supremum is longer but more transparent about why the limit exists at all.

Where the recognition trips up

The pattern fires on any question of the form "does this infinite-digit decimal equal this finite thing?" — but only those. Watch the adjacent traps.

Not every infinite decimal is repeating. \pi = 3.14159\ldots is infinite and non-repeating; no finite closed form exists. The pattern here does not apply — you cannot sum \pi's digits as a geometric series because the digits do not repeat with a common ratio. If the question asks whether a non-repeating decimal equals a fraction, the answer is almost always no, and the proof goes the other way (show the fraction has a terminating or repeating decimal, contradicting the assumption). See Non-terminating vs Non-repeating Decimals.

The geometric series formula requires |r| < 1. It always does for a decimal expansion (r = 10^{-k} for block length k), but if a disguised problem gives you a series with |r| \ge 1, the formula fails and the sum diverges. Check the ratio before quoting the formula.

"Equals" is literal, not "approximately equals". The most common confusion is still that 0.999\ldots is "just below" 1. It is not — it is 1. The two decimal strings are two different names for the same real number. The article Why 0.999... Feels Slightly Less Than 1 walks through the intuition gap. The nested-intervals visualisation at 0.999... = 1 as Nested Intervals Shrinking to a Point makes the collapse visible.

The drill

When you see any of these prompts, both routes should appear in your head within five seconds:

All five collapse to: identify the truncations → supremum, or identify the geometric sum → closed formula. Pick whichever gives you a shorter path. For short blocks (\overline{9}, \overline{3}), either works. For long blocks (\overline{142857}), prefer the series. For a proof-based setting where foundations matter, prefer the supremum.

The recognition — not the computation — is the skill. Once the problem has been slotted into the right template, the rest is algebra.