Pi & Euler's number: decimal places, records and number experiments

Pi begins with 3.14159, Euler's number with 2.71828. Both decimal expansions continue indefinitely. Both numbers have exact definitions, so you can calculate with them without writing out every decimal place.

You might look up the decimal places of pi to copy a set number of digits, search for a pattern or make sense of a record. With e, the question is usually why this number appears in growth models. The two constants have different uses:

To try them directly, calculate decimal places of pi or calculate decimal places of Euler's number. The tools show finite segments, not the complete decimal expansions.

Understanding pi: a finite number with infinitely many digits

For every circle in the Euclidean plane, π = C / d, where C is the circumference and d is the diameter. A larger circle has a larger circumference, but the ratio stays the same. With radius r, the familiar formulas are C = 2πr and A = πr² for the area.

Pi is not a large number, and more accurate measurements do not change its value. The number lies between 3 and 4. It is the decimal expansion that is infinitely long. A measurement with string and a ruler can illustrate the circle ratio, but its accuracy is limited by the measuring equipment.

The first 100 decimal places of pi

The following two lines belong together. Each contains 50 decimal places, with the leading 3 added before them. The segment is truncated, not rounded:

3.14159265358979323846264338327950288419716939937510
  58209749445923078164062862089986280348253421170679

This sequence has been checked against the District of Columbia digit sheet. To find the 100th decimal place, count only to the right of the decimal point. Spaces and line breaks are not part of the number. Following this convention helps you avoid mistakes when comparing digit lists.

A formula can describe pi exactly even though its decimal result can never be written out in full. Infinite series are also exact representations when they refer to their limits. A calculation stopped after a finite number of terms, however, gives an approximation. "Cannot be expressed as a fraction" does not mean "has no exact formula".

Euler's number: why e appears in growth

Euler's number is the base of the natural exponential function e^x and the natural logarithm ln(x). Its first 50 decimal places, again truncated, are:

2.71828182845904523536028747135266249775724709369995

These opening digits match the table in OEIS entry A001113. The repeated 1828 near the start is a short digit block, not a continuing period. You can calculate more digits with the e decimal places tool.

From compound interest to a limit

To see where e comes from, suppose you start with 1,000 euros at a nominal annual interest rate of 100 percent for one year. This deliberately high rate serves only to explain the idea. If interest is credited several times, divide the nominal rate evenly between those periods; each interest payment then earns interest itself.

Round only at the end of this calculation. Crediting interest more often increases the final amount, but it still approaches a finite limit. The ratio of final capital to initial capital satisfies e = lim(n→∞) (1 + 1/n)^n. Actual bank products may have different interest agreements and rounding rules.

The general continuous model is K(t) = K₀ × e^(rt). K₀ is the initial capital, r the continuous annual rate as a decimal and t the time in years. Enter a rate of 5 percent as 0.05. The percentage calculator helps with percentage problems, but does not calculate this entire interest model. Do not equate the nominal rate, effective annual rate and continuous rate without checking them.

Approximating e yourself with a series

A second approach needs no interest model: e = 1/0! + 1/1! + 1/2! + 1/3! + …. The exclamation mark denotes the factorial. For example, 3! is the product 1 × 2 × 3. By definition, 0! equals 1. The series therefore starts with 1 + 1 + 1/2 + 1/6 and adds progressively smaller positive terms.

Every finite partial sum is a rational number. Their limit, though, is irrational. There is no contradiction: a property of an individual finite calculation need not hold for its limit. An irrational number can also be approximated by increasingly long finite decimals.

To know how accurate your approximation is, you need more than a sum of a few terms. You also need to know how much the omitted remainder can contribute. Adding the terms from 1/0! up to and including 1/9! gives the separately verified exact partial sum S₉ = 98,641/36,288 ≈ 2.718281525573.

The first missing term is 1/10!. Each subsequent term is at most one eleventh of its predecessor. A geometric series therefore gives the upper remainder bound 0 < e − S₉ < 11/(10 × 10!). The missing remainder is less than 0.000000304.

This gives the narrow interval 2.718281525573 < e < 2.718281828704. Both bounds round to 2.71828 at five decimal places. You can therefore justify this rounding without trusting a long digit file. This bounds the error rather than just producing plausible-looking digits.

What makes this base natural

For the function e^x, the instantaneous slope at every point equals the function value: f′(x) = e^x. This makes e^x a simple way to describe proportional change. In N(t) = N₀ × e^(kt), a positive k represents growth and a negative k represents decay.

This is a modelling assumption, not a promise of unlimited growth. Food and habitat can limit a population's development. An exponential model may fit that population only over a limited period. The presence of e in a formula does not automatically make its underlying assumptions correct.

Both e and pi appear in statistics: the probability density of the standard normal distribution is f(x) = e^(−x²/2) / √(2π). To understand the bell curve, its properties matter more than memorising long digit lists.

Irrational, transcendental and normal: three different statements

These terms answer different questions. Keeping them separate helps explain many apparent surprises in digit sequences.

Rational means that a number can be written as a ratio of integers with a nonzero denominator. Its decimal expansion terminates or becomes periodic. For example, 1/8 is exactly 0.125, while the 3 in 1/3 repeats indefinitely. A finite decimal is also rational.

Irrational means that no such fraction exists. The decimal expansion neither terminates nor becomes permanently periodic from some point onwards. Pi and e are irrational. Individual digits and short blocks can still repeat. Without this distinction, the statement "pi never repeats" is false.

Transcendental means that the number is not a root of any nonzero polynomial with integer coefficients. Pi and e also have this stronger property. Irrationality alone is not enough: √2 is irrational, but satisfies the equation x² − 2 = 0.

A number would be normal in base 10 if, in the long run, every finite digit block of a given length reached its expected frequency in the decimal expansion. Each individual digit would have a limiting share of 10 percent, and each particular two-digit block a share of 1 percent. This does not require exactly equal counts in every short segment.

Normality remains unproven for pi and e. However large a calculated sample is, it remains finite. It can support a conjecture, but cannot prove a statement about every digit. Even equally frequent individual digits would not suffice, because normality also concerns all longer blocks.

Why disordered does not mean random

The decimal digits of the constants are fixed. A correct calculation produces the same segment again. So these digits are neither secret nor random, and you should not use them as cryptographic key material.

For comparison, use the random number tool to generate an integer between 0 and 9 several times, then note the results. This tool uses Math.random(), a pseudorandom generator without cryptographic security. It suits a classroom experiment, not passwords or security-critical draws. Similarity between two digit sequences proves neither randomness nor normality.

Reading a small frequency experiment correctly

Take the first 100 decimal places of pi and make a tally for each digit from 0 to 9. Repeat the count with a longer segment. Record the sample size and the proportion of each digit in each case. This lets you compare relative frequencies rather than just absolute counts.

In a number that is normal in base 10, these proportions approach 10 percent in the long run. For any particular finite segment, this implies neither a guaranteed deviation bound nor a prescribed count of each digit. A striking cluster in a short list is still just an observation.

A second question helps when evaluating the result: did you count only individual digits, or also blocks such as 00 and 01? A list can look balanced for individual digits and still have clear patterns in longer blocks. A balanced tally alone therefore cannot establish normality. The experiment examines your segment, not the entire infinite expansion.

The Feynman point and birthday searches without false promises

The Feynman point is the sequence 999999 starting at the 762nd decimal place of pi. The six nines occupy positions 762 to 767. The leading 3 is not counted. This finite block does not contradict irrationality: only a period that repeats permanently from some point onwards would be incompatible with it.

To check the block yourself:

1. Calculate at least 767 decimal places with the pi tool, or 1,000 for easier comparison.
2. Look only at the digits to the right of the decimal separator. Remove any inserted spaces and line breaks when transferring them.
3. Search for 999999 and check its starting position by counting from 1.
4. Record whether your text search reports positions or character indices. Counting from 0 changes the numerical index.

A birthday search also needs a fixed format. Day and month, a two-digit year or a full year produce different search sequences. Leading zeros are part of the sequence. Unless you specify the format, others will struggle to reproduce a claimed match.

A missing match in 100,000 digits means only that the chosen sequence was not found in that segment. It does not prove that it is absent later. Conversely, a single match says nothing about the normality of the entire number. The pi tool supplies the digits; this does not imply an automatic date-search feature.

For the record of 314 trillion digits, Guinness reports that the dataset contains every possible combination of 13 digits. That is what Guinness reports finding in a finite dataset, not a proof of normality. A claim about every sequence of any length would be much stronger and cannot be deduced from it.

Decimal digit records: what the large numbers tell us

Sources re-checked: 19 October 2026. Guinness lists 314,000,000,000,000 digits for pi. The central record list maintained by the y-cruncher developer lists 35,000,000,000,000 digits for e. Both entries agree with the research findings from 10 October. Publicly documented records may lag behind calculations that have not yet been published.

The English trillion corresponds to the German Billion. The pi record is therefore 314 trillion, or 314 Billionen in German, not 314 Trillionen. Confusing these terms when translating an English report into German gives the wrong order of magnitude.

The sources differ on the date of the latest pi calculation: Guinness gives 18 November, while the y-cruncher table gives 19 November 2025 as the completion date. The reason is unclear. Do not treat the record date, completion date and a later announcement as interchangeable.

These runs mainly test powerful calculation methods, how storage handles input and output, and system stability. The StorageReview report describes a single server with 40 SSDs of 61.44 TB each and 1.5 TB of RAM for the pi record. These are operator-reported specifications, not an independent assessment of costs or energy use.

The e entry separately lists 94.5 hours for calculation and 92.5 hours for verification. These figures are not a single overall runtime. For e, the record list requires, among other things, matching results from different algorithms. We checked the source, but did not recalculate the complete dataset.

How many digits are useful?

The accuracy you need depends on the task. For a circle calculation, the measurements and the permitted error matter, not the number of available digits of pi. A thousand extra decimal places cannot make a diameter measured to the nearest millimetre more precise.

In a checked example, a circle with a diameter of exactly one metre has a calculated circumference of 3.14 metres when you use π ≈ 3.14. This falls short of the mathematical value by around 1.59 millimetres. Whether that is acceptable depends on the application. A rough sketch has different requirements from a numerical method with a strict error bound.

Truncation and rounding are different operations. Truncation removes all subsequent digits. When rounding, the next digit determines whether the last retained digit increases. A correct prefix is therefore not automatically the correctly rounded value of the same length.

The calculation method must also support the number of digits you need. Appending extra characters does not give an ordinary floating-point number extra precision. The pi tool uses a spigot algorithm for up to 100,000 decimal places, while the e tool uses a series for up to 10,000 decimal places. Do not confuse these limits with record-scale calculations.

Before copying a digit list, run through this short checklist:

• Is the requested length defined as decimal places or as the total number of digits?
• Is this a prefix or a rounded approximation?
• Does a reported match count from 0 or from 1, and does it include the digit before the decimal point?
• Will leading zeros and the digit order survive pasting?
• Does the calculation's precision match the input data?

Where the constants come from: selected historical milestones

Pi was known long before its current symbol. Archimedes bounded its value between 223/71 and 22/7 using inscribed and circumscribed polygons. This was a method with bounds, not just a circumference measurement. Around 1400, the Indian Kerala school associated with Madhava developed infinite series for trigonometric functions and pi. Much of the surviving account comes from reports by his successors.

William Jones used the symbol π in its modern sense in 1706; Euler adopted it later. Lambert proved irrationality in 1761, Lindemann transcendence in 1882. These are different results. Transcendence implies, in particular, that the classical squaring of the circle with compasses and a straightedge is impossible.

For e, the historical starting point is compound interest. In 1683, Jacob Bernoulli studied the limit now written as (1 + 1/n)^n. Euler treated e systematically in his "Introductio in analysin infinitorum", published in 1748. His work on the irrationality proof was written in 1737 and published in 1744. Hermite proved transcendence in 1873.

Naming the number after Euler does not explain why he chose the letter e. There is no established account showing that he chose e as his own initial. To understand the mathematics, you only need to keep the number, its symbol and its later name separate.

Number experiments with fractions, primes and Fibonacci

You can compare decimal places with other ways of writing numbers. Try these tasks with small values so you can follow the results.

Simplifying a fractional approximation does not improve its accuracy

Use simplify fractions to reduce 44/14 to 22/7. Dividing the numerator and denominator by their greatest common divisor leaves the value unchanged. Both fractions approximate pi equally well.

In the fraction to decimal tool, 22/7 appears rounded as 3.142857. The calculator displays at most six decimal places and uses a decimal point. The actual decimal expansion is periodic. A short display therefore does not prove that a number has finitely many decimal places.

Using the denominator to tell whether a decimal terminates

For a fully reduced fraction with a positive denominator, its decimal expansion terminates if and only if the denominator contains no prime factors other than 2 and 5. For 1/8, the denominator is 8 = 2³ and the result is exactly 0.125. For 1/3, the denominator has a prime factor other than 2 or 5, namely 3, so the expansion is periodic.

You can examine small denominators with prime factorisation. Reduce the fraction first, otherwise common factors can distort the test result. This rule concerns rational numbers and does not turn either pi or e into a ratio of integers.

Checking Fibonacci numbers for primality

Generate a short list with the Fibonacci sequence tool. The input determines the number of terms, not an upper value limit. The sequence begins with 0, 1, 1, 2, 3, 5; each subsequent term is the sum of the two preceding terms.

Then check 89 and 144 with the prime checker: 89 is prime, 144 is not. The factorisation of 144 is 2 × 2 × 2 × 2 × 3 × 3. Checking these integers tells you nothing about whether pi or e are prime. Prime numbers are integers greater than 1; these constants do not belong to that category.

Building a provably normal number from primes

Display the first primes with the prime number list and concatenate them after the decimal point. This produces the beginning 0.23571113171923…. Continuing this construction indefinitely gives the Copeland-Erdős constant.

Its normality in base 10 has been proven, unlike that of pi and e. An experiment with a finite list provides the beginning of this number, not the proof. It does, however, show why "normal", "computable" and "randomly generated" are different properties.

The connection between pi and e

Euler's identity e^(iπ) + 1 = 0 connects the two constants exactly. Here i is the imaginary unit, with i² = −1. The equation follows from the complex exponential function, not from a chance coincidence of their decimal places.

A geometric interpretation uses rotations: in e^(iθ), θ describes an angle in radians. At θ = π, you reach the point −1 on the unit circle. Adding 1 gives 0. If you have only encountered real exponents so far, this requires extending the number system.

Start with a small experiment: calculate a segment of pi, check the Feynman point or compare a fraction with its rounded decimal display. Always record what you actually examined: a finite segment, an approximation or a proven property of the number.