ROUNDSIG Function (LibreOffice Calc)

=ROUNDSIG(Value; Digits)

There are exactly two required arguments, without optional defaults. Digits=3 does not mean three digits after the decimal point. The ; separating arguments follows the English-language Calc formulas documented by LibreOffice Help; localized settings may change separators, names, number formats, or decimal punctuation. For portable sharing, check the recipient’s application and locale rather than assuming semicolons and decimal points will paste unchanged.

The precise official formulation and the requirement that Digits be a positive integer are from LibreOffice Calc Help: ROUNDSIG, checked 2026-10-13.

Try it immediately

Enter the following numbers as numbers, not as quoted text. The formulas can be placed in column C.

Each underlying literal equivalent was executed with Calc 25.2.3.2. Cell-reference variants use the same documented argument form; the independent assertion set was built primarily from numeric literals, so the table’s cell-based instructions should not be mistaken for additional separately counted native assertions. A returned numerical 123500000 could display as 1.235E+08 when the cell uses a scientific-number format. Formatting is not a change in the calculated value.

For an ordinary nonzero decimal number, count from its first nonzero digit and then count subsequent digits. Leading zeros merely locate the decimal point; they are not counted. Zeros between meaningful digits do occupy positions. For example, 0.000123456789 has first significant digit 1, not the zeros before it. To retain five significant digits, ROUNDSIG uses the sequence 1,2,3,4,5 and looks at the next digit when rounding.

This differs from ROUND, which counts decimal places relative to the decimal separator. For instance, =ROUNDSIG(123.456789;3) rounds to approximately 123, while =ROUND(123.456789;3) would target three digits after the decimal point instead. The latter comparison is the mathematical distinction specified by the functions; this batch did not include an independent native assertion for the ROUND expression.

Important: Excel-style or scientific textual formatting such as displaying 120000000000000 as 1.2E14 is a display decision. It does not prove that trailing zeros are measured, verified, or physically meaningful. If laboratory precision must be communicated with explicit trailing zeros, carefully select a number format or use a string representation, documenting that such formatting is presentation only.

A conceptual way to understand the operation for finite, nonzero x and positive integer n is:

1. Find the power of ten of the first significant digit: k = FLOOR(LOG10(ABS(x))).
2. The intended decimal rounding-place index is n - 1 - k.
3. Round x to that many places, including negative place counts for hundreds, thousands, and so on.

For example, 12345.67 begins at 10^4. Keeping three significant digits corresponds to rounding at the hundreds place, producing 12300. For 0.0098765, its first significant digit is at 10^-3: keeping two digits rounds at the fourth decimal place, producing 0.0099.

A conceptual comparison, for nonzero values in a suitable ordinary numeric range, is:

=ROUND(A2;B2-1-INT(LOG10(ABS(A2))))

That is an explanatory, not natively tested in this batch, expression. Use IF(A2=0;0;...) to guard LOG10(0) if adapting it. Rounding behavior at extreme magnitudes, midway values, and floating-point boundaries can differ from a built-in function; test the specific workbook. There is no need to replace native ROUNDSIG with this longer formula in Calc.

Imagine readings in millimetres arrive at very different sizes. Place the unrounded values in column B and the number of requested significant figures in column C. In D2 use:

=ROUNDSIG(B2;C2)

Copy down only after confirming that every C value is a positive integer. For a constant report rule, use =ROUNDSIG(B2;3) instead. Examples from the native numerical fixture:

Practical caution: apply significant-figure rounding at a defined reporting boundary, not repeatedly inside every intermediate calculation, because rounding at each stage can compound errors. Keep raw observations in one column and derived display/report values in another. This is a workflow recommendation, not a guarantee of metrological uncertainty compliance.

Midway examples can distinguish rounding conventions. In Calc 25.2.3.2 native tests, =ROUNDSIG(2.5;1) returned 3, and =ROUNDSIG(-2.5;1) returned -3; =ROUNDSIG(1005;3) returned 1010, with the negative counterpart yielding -1010. These observations are version-specific. The official ROUNDSIG page describes significant-digit rounding, but does not provide an exhaustive specification of every binary-float tie case. Do not infer behavior for all versions or for values whose entered decimal representation is not exactly representable in binary.

1. Scientific and engineering notation

Instead of estimating the number of decimal places for each row, use ROUNDSIG to put measurements on a comparable precision scale. =ROUNDSIG(123456789012345;2) returned 120000000000000 in the tested Calc; a scientific display format can represent it as 1.2E14. The displayed exponent does not change the underlying stored number.

2. Precision controlled by a data column

Store validated integer precision settings in C2:C100. Use =ROUNDSIG(B2;C2), and apply data validation (whole numbers greater than zero) to the settings column. This avoids scattered hard-coded rounding rules when different measurements need different reporting precision.

3. Comparing an original calculation with its published figure

Keep B2 as the full-precision computation; put =ROUNDSIG(B2;3) into C2. When auditing, report both so differences caused by presentation rounding remain visible. In a financial or billing workflow, use the legally or contractually required decimal/monetary rounding rule instead of casually substituting significant figures.

4. Guard against invalid precision settings

A defensive pattern for nonnumeric or invalid settings is conceptually:

=IF(ISNUMBER(C2);IF(AND(C2>=1;C2=INT(C2));ROUNDSIG(B2;C2);"Check Digits");"Check Digits")

This formula has not been separately executed in this batch, so treat it as an adaptation example rather than a native-verified output. It expects numeric B2. The outer IF rejects nonnumeric C2 before evaluating INT; upstream errors and the value being rounded still need their own validation. INT and logical functions may be localized or use different separators depending on locale.

Documentation-versus-execution nuance: LibreOffice 26.8 Help states that Digits must be an integer greater than zero. However, the 25.2.3.2 native diagnostic =ROUNDSIG(123.45;1.5) returned 100, rather than an error. Similarly, numeric-looking string digits and TRUE() were coerced in that build. These are observations, not supported usage recommendations; they do not override the documented requirement. Enter clean positive integers for predictable files.

In native tests =ROUNDSIG(0;4) returned 0; no logarithm guard is needed for the built-in function. Very small =ROUNDSIG(1E-300;2) returned 1E-300 in the observed Calc build. =ROUNDSIG(9.99999999999999E99;3) returned approximately 1E+100 after rounding. These observations do not establish support for unlimited exponent size or arbitrary decimal precision: Calc stores numerical values within implementation limits. Requesting many significant digits cannot recover information absent from the input’s stored floating-point value.

According to LibreOffice’s official ROUNDSIG Help checked 2026-10-13, this function has been available since LibreOffice 5.4. The same page states that it was not included in OpenDocument OpenFormula 1.3 and identifies its extension namespace as ORG.LIBREOFFICE.ROUNDSIG. That namespace matters when exchanging formula-containing files: availability in the LibreOffice engine is not a guarantee that another program will parse, retain, or recalculate the same expression.

• ROUND — fixed decimal places
• ROUNDUP — round away from zero
• ROUNDDOWN — round toward zero
• MROUND — round to a multiple
• ABS — magnitude for input-scale inspection
• LOG10 — base-ten logarithm for a portable approximation
• IF — handle invalid input settings

1. Official function definition, syntax, four official examples, 5.4 introduction, extension namespace, ODF 1.3 exclusion: LibreOffice 26.8 Help — ROUNDSIG. Retrieved/reviewed 2026-10-13.
2. Reproducible numerical and diagnostic cases: saved dated results, ODS and calculation script. Formula engine: LibreOffice Calc 25.2.3.2 Linux, executed through UNO on 2026-10-13. Numeric assertions based on independently calculated Decimal significant-figure rounding; actual Calc values recorded separately.