LOGNORM.INV in ONLYOFFICE: Lognormal Quantiles from Log-Scale Parameters

Create DurationModel. Put Parameter in A1 and Value in B1, then enter these values and one formula:

This is an assumed lognormal model for a fictional task duration, not observed data. If X is the duration measured numerically in minutes, ln(X) is modeled as normal with mean μ and standard deviation σ. Use natural logarithms, not base-10 logarithms. The numerical unit must stay consistent when transforming data.

Here exp(μ)=10 is the median, not the arithmetic mean duration. The arithmetic mean is exp(μ+σ²/2), approximately 11.3314845306683 minutes; the mode is exp(μ−σ²), approximately 7.78800783071405. Entering 10 as the mean argument instead of LN(10) specifies a radically different model. A positive skew does not by itself prove real data follow a lognormal distribution.

The official reference takes probability, mean and standard_dev. Probability must be strictly between 0 and 1, and the logarithmic standard deviation must be positive. The returned threshold is in the original units of X, not the units of ln(X).

The mathematical transformation is exp(μ+σΦ⁻¹(p)). At p=0.5 the standard normal quantile is zero, so the result is exp(μ), the median. An ordinary arithmetic mean of positive observations is not a valid substitute for μ without the appropriate model conversion.

D4 and D2 enclose the central 90% of the assumed distribution. D5 and D6 enclose 95%. The intervals are not symmetric in minutes: equal tail areas become symmetric on the logarithmic scale, not the original scale. Using 0.95 for the upper bound and 0.05 for the lower bound does not create a central 95% interval.

In F2 enter =NORM.S.INV(B7) for approximately 1.64485362695147. In F3 enter =EXP(B2+B3*F2) for the D2 target. Then =LOGNORM.DIST(D2,B2,B3,TRUE) has mathematical target approximately 0.95. Treat a round trip as a tolerance-based check, not exact floating-point equality.

The paired 5th and 95th percentiles have product exp(2μ)=100 in this model. Their distances from 10 are unequal, but their ratios to 10 are reciprocals. This is a useful check for confusing additive normal symmetry with multiplicative lognormal symmetry.

For a one-sided upper-tail allowance of 5%, enter p=0.95. Supplying p=0.05 instead returns the low threshold. These are assumed duration percentiles, not a promise that a future schedule will finish on time and not a confidence interval for an estimated parameter.

The tagged source rejects p≤0, p≥1 and σ≤0 with #NUM!. The mathematical lower-endpoint limit is 0 and the upper-endpoint limit is unbounded, but =LOGNORM.INV(0,LN(10),0.5) and the corresponding p=1 call are source-predicted errors. Do not substitute an error result for either limit.

LOGINV uses the same scalar mathematical expression, but is not implemented by inheritance. It scalarizes explicit array arguments; LOGNORM.INV passes through array-oriented argument helpers. Separate implementations and error-handling paths mean a scalar identity is not evidence of complete formula or array equivalence. No native array workflow is claimed.

This inverse uses a normal-quantile approximation followed by exponentiation; it is not an iterative solver fitting μ and σ to observed data. Very large exponents can overflow, very negative ones can underflow, and probabilities rounded to 0 or 1 hit rejected endpoints. The numeric wrapper rejects nonfinite values, so do not promise finite outputs for every mathematically valid parameter combination. Use moderate examples and verify exact-build behavior when scale or tail precision matters.

Checked October 18, 2026 against current official help and SDKJS tag v10.0.0.169, commit 3679e0375a758d7dc59d31337a45461be3525ed8. Introduction dates and a final supported-version endpoint are not established. Desktop Editors, Docs, mobile releases and imported files require separate compatibility checks.

Neither the ONLYOFFICE application nor its calculation engine was executed for this guide. The numerical targets are independently calculated mathematical values; code-path and error observations are manual source traces. They are not screenshots, displayed workbook results or promises of identical final digits. Use English function names and comma separators as shown, adapting separators to the editor’s locale.

Keep inputs and intermediate probabilities numeric and unrounded. Format the display rather than replacing values with rounded strings. These moderate scalar examples do not certify extreme-tail accuracy, unusual type coercion, array entry or dynamic spill. Model assumptions must be checked separately from arithmetic correctness.

• Official LOGNORM.INV reference, checked October 18, 2026.
• Pinned implementation.
• Pinned normal helpers and beta helper.
• LOGNORM.DIST.
• LOGNORMDIST.
• LOGINV.
• NORM.S.INV and NORM.S.DIST.