Why does engineering notation require an exponent that is a multiple of three?

The constraint aligns the exponent with SI prefixes such as milli (10⁻³), micro (10⁻⁶), kilo (10³), and mega (10⁶), all of which are powers whose exponent is already a multiple of three. That alignment lets you swap the exponent for a prefix directly — for example, 4.7e-6 farads reads as 4.7 microfarads without further arithmetic.

Does 1e-5 become 10e-6 or 0.01e-3 in engineering notation?

It becomes 10e-6. The exponent floors down to the next lower multiple of three, which for −5 is −6, and the one-digit shift moves into the coefficient. Integer division toward zero would give the wrong answer, which is why the converter explicitly floors the exponent even for negative values.

Will large integers lose precision when converted to engineering notation?

No. The coefficient digits are joined as text and shifted by string position rather than passed through JavaScript's Number type, so values larger than Number.MAX_SAFE_INTEGER keep every nonzero digit. The exact value 9007199254740993 comes out identically in scientific, engineering, and complete decimal form.

What happens when the full decimal expansion is extremely long?

The complete decimal output is capped at 100,000 code units. If the exact expansion would exceed that budget, the conversion returns a single error and shows none of the three forms, rather than truncating the result or substituting an ellipsis. The input limit of 20,000 code units and coefficient limit of 10,000 digits bound the worst-case work.

How to Convert Scientific Notation to Engineering Notation

Engineering notation expresses a number as a coefficient times a power of ten whose exponent is restricted to multiples of three — 0, ±3, ±6, ±9, and so on — which leaves one, two, or three digits before the decimal point for any nonzero value. That single constraint is what separates engineering notation from ordinary scientific notation: in scientific notation, 12345 becomes 1.2345 × 10⁴ with the exponent 4, but in engineering notation the same number becomes 12.345 × 10³ because the exponent must drop to the next lower multiple of three. The conversion rule, then, is simply to take any value written in scientific form, floor its exponent to the next lower multiple of three, and shift the freed-up digits back into the coefficient. For very small numbers, the rule works in reverse: 1 × 10⁻⁵ floors to 10⁻⁶, giving 10 × 10⁻⁶, not 0.01 × 10⁻³. Getting that floor direction right is the most common source of off-by-three errors when converting by hand.

How Engineering Notation Differs From Scientific Notation

Both notations write a number as a coefficient × 10^exponent. The only structural difference is the rule applied to the exponent. Scientific notation requires exactly one nonzero digit before the decimal point and lets the exponent be any integer. Engineering notation requires the exponent to be a multiple of three, which lets the coefficient grow to one, two, or three digits before the decimal point for nonzero values. That choice is not arbitrary — it is exactly what makes engineering notation line up with SI prefixes. Micro (10⁻⁶), milli (10⁻³), kilo (10³), mega (10⁶), giga (10⁹), and tera (10¹²) are all powers whose exponent is a multiple of three, so a value already in engineering form can be relabeled with a prefix by simply reading the exponent. A datasheet that lists 4.7 × 10⁻⁶ farads is documenting the same capacitance as 4.7 µF; the latter is engineering notation with the prefix micro replacing the exponent.

Property	Scientific notation	Engineering notation
Exponent allowed	Any integer	Multiple of three only
Digits before the decimal point	Exactly 1 (nonzero)	1, 2, or 3 (nonzero)
Example for 12,345	1.2345 × 10⁴	12.345 × 10³
Example for 0.000047	4.7 × 10⁻⁵	47 × 10⁻⁶
Aligns directly with SI prefixes	Indirectly	Directly

The exponent rule matters even when the prefixes are not in play. Spreadsheets, calculators, and IEEE 754 displays all benefit from a single fixed-width coefficient. Physics labs, electronics workbenches, and metrology reports consistently prefer engineering form because it produces a coefficient with at most three digits, which is short enough to write in a table column or print on a label without breaking layout. NIST's Guide for the Use of the SI treats the alignment of notation with prefix-labeled units as the reason engineering form is preferred in laboratory and engineering contexts.

Converting Scientific Notation to Engineering Notation

Open the Scientific Notation Converter and follow these steps:

Enter a value in the input box. Type an ordinary decimal such as 12345, 0.00123, or .5, or paste an e/E scientific value such as -4.2e-7. Use a point as the decimal separator, leave out commas and units, and include at most one optional leading sign.
Select Convert notation. The converter parses the input in your browser using a strict grammar that accepts an optional sign, decimal digits with one optional decimal point, and one optional signed whole-number exponent. Anything else — commas, spaces, hex digits, repeated points, or a bare decimal point — is rejected.
Read the engineering row. All three output forms appear together: the normalized scientific form, the engineering form, and the complete decimal value. The engineering row is the one whose exponent ends in a multiple of three.
Copy the exact form you need. Each output has its own Copy button so the clipboard never gets the wrong format. The engineering form is the right choice for datasheets, prefix labeling, or code that expects powers of ten aligned with SI.
Note the explicit limits. Raw input is capped at 20,000 UTF-16 code units, the coefficient at 10,000 decimal digits, and the complete decimal output at 100,000 code units. If a result would push past any of those budgets, the converter returns one error instead of a partial or rounded value.

Worked Example: 12345 in Three Forms

Take the integer 12345. Writing it in normalized scientific notation puts one nonzero digit before the decimal point:

12345 = 1.2345 × 10⁴, or 1.2345e+4 in compact e notation.

To convert that to engineering notation, floor the exponent 4 down to the next lower multiple of three. The next lower multiple is 3, a drop of one step. Shift one digit from the exponent back into the coefficient:

12345 = 12.345 × 10³, or 12.345e+3.

The complete decimal form is just 12345 — no fractional zeros required. If you want to confirm the rule by hand, count digits: the coefficient 12.345 has two digits before the point (12) and three after the decimal (345), and the exponent 3 is a multiple of three. The same value passes through the converter and the engineering row comes back as 12.345e+3 every time, because the tool feeds every output back through its own normalizer and confirms that the result is canonical before showing it. For comparison, Python's Decimal.to_eng_string uses the same floor-down rule for the same reason.

Input Formats the Strict Parser Accepts

The converter deliberately rejects anything ambiguous, so it is worth knowing the exact grammar the parser enforces before you paste a value.

Accepted signs. A leading + or − is optional. −0 and +0 are both accepted and then canonicalized to plain 0.
Accepted coefficient structure. Decimal digits with at most one decimal point. A leading decimal point is fine when digits follow it, so .5 is valid. A trailing decimal point after digits is also fine, so 123. is valid. A bare . with no digits on either side is not.
Accepted exponent. One optional e or E followed by a signed whole number. The exponent can be far larger than JavaScript's safe-integer range because it is parsed with BigInt before any arithmetic happens.
Whitespace. Leading and trailing whitespace are trimmed; embedded spaces inside the number are rejected.

Rejected formats. Thousand separators (1,000), space separators (1 000), hexadecimal or other base prefixes, the strings Infinity and NaN, a coefficient with two or more decimal points, an exponent with no digits (1e), and a coefficient with both a leading and trailing point are all rejected by the parser. When rejection happens the converter shows a single error and does not attempt a partial conversion. That strictness is what lets it guarantee every nonzero coefficient digit is preserved exactly — see Wolfram MathWorld on scientific notation for the conventions the strict grammar implements.

Limits, Edge Cases, and Why Zero Canonicalizes

The output side of the converter is governed by three explicit budgets: raw input up to 20,000 UTF-16 code units, coefficient up to 10,000 decimal digits, and complete decimal output up to 100,000 code units. None of these caps silently round or truncate. If a conversion would require even one more code unit than the budget allows, the entire conversion fails with one clear error — the three-part output is not shown at all, so a reader cannot mistake a truncated result for the canonical value.

Two edge cases deserve attention because they trip up manual conversions. First, negative exponents floor the same way positive ones do: 1e-5 floors to exponent -6 (the next lower multiple of three), giving 10e-6, not 0.01e-3. The sign belongs to the value, not to the exponent, so -4.2e-7 floors to exponent -9 and yields -420e-9. Second, zero has no unique normalized exponent, so the converter canonicalizes every zero — including -0, 0.000, and 0e999999 — to the single form 0e+0 with decimal 0. Leading zeros, trailing zeros, an explicit plus sign on the input, and a negative sign on zero are all treated as cosmetic rather than semantic, which means the same numeric value comes out the same way no matter how it was formatted.

Large integers are preserved exactly even when they exceed Number.MAX_SAFE_INTEGER. The coefficient never enters JavaScript's floating-point Number, so 9007199254740993 round-trips back as 9007199254740993, including in its engineering form. If you need to confirm a long value against a source measurement, run it through the Scientific Notation Converter and compare the engineering row against the original. The converter is a notation tool rather than a rounding or significant-figures tool, so retain your measurement record separately when trailing zeros carry precision meaning.