PROB Function (Gnumeric)

• Introduced: First release not established; present in the retained 1.12.57 source.
• Supported in: Gnumeric 1.12.62; no retirement is established.
• Scope: Desktop Gnumeric with its Statistics plugin. English comma-separated formulas are shown.
• Checked: October 14, 2026. The tables and named advanced check use 1.12.57 ssconvert. Current 1.12.62 source was inspected; that release was not executed.

PROB sums supplied probability masses whose associated outcomes lie in a closed interval. Omitting upper_limit asks for the mass at exactly lower_limit. The function does not estimate probabilities from frequencies or normalize invalid weights automatically. Versioned function help.

=PROB(x_range,prob_range,lower_limit[,upper_limit])

• x_range: outcome values.
• prob_range: corresponding strictly positive probability masses, each no greater than 1. Zero masses are rejected in the tested and current implementations.
• lower_limit: inclusive lower bound.
• upper_limit: optional inclusive upper bound; defaults to lower_limit. Reversed bounds are not swapped.

Brackets mark optional arguments; do not type them. Current implementation.

Enter J2:J5 as 0,1,2,3 and K2:K5 as 0.1,0.2,0.3,0.4. Diagnostic probability columns are:

U2:U5 contains number 0, text skip, number 1, number 2. V2:V5 contains number 0.2, number 0.3, text skip, number 0.5. F7 is genuinely empty. Ordinary native outputs are rounded to 12 decimals; the tolerance examples retain their additional digits.

Use outcomes 0,1,2,3 in J2:J5 and probabilities 0.1,0.2,0.3,0.4 in K2:K5. =PROB(J2:J5,K2:K5,1,2) gives 0.5. Omitting the upper bound gives the single-outcome mass 0.2 at x=1.

A reversed request from 2 down to 1 returns zero after validation. Decide the interval explicitly; do not assume a compatibility application will use the same reversal rule. The native sum of disjoint intervals 0–1 and 2–3 is 1.

The sum check accepts floating-point differences within n × 2 × machine_epsilon; for this binary64 build epsilon is 2^−52. Four probabilities summing approximately to 1.000000000000001 pass and retain that sum, while the tested approximately 1.00000000000001 set fails. This tolerance absorbs tiny arithmetic errors and is not permission to use unnormalized weights.

Keep both ranges purely numeric and aligned. Each is filtered independently. In the diagnostic U/V columns, outcomes 0,text skip,1,2 and probabilities 0.2,0.3,text skip,0.5 become numeric lists 0,1,2 and 0.2,0.3,0.5. PROB then assigns mass 0.3 to outcome 1, although those entries did not share an original row. Equal retained counts do not guarantee correct pairing.

The relevant argument guards and calculation sequence are preserved in the retained 1.12.57 and 1.12.62 callbacks. This is a source comparison; later-release output was not tested.

A zero, negative or greater-than-one probability gives #NUM!, as does a sum outside the tolerance. Unequal retained counts give #N/A. Empty ranges fail the sum check. Clean or reject incomplete pairs together before calling PROB; do not interpret the independently compacted diagnostic result as a valid model of the original rows.

• Current worksheet implementation
• Tested-version worksheet implementation

Both the stale SUM cache and the uncached SUM recalculated correctly. Saved formula identities and typed inputs were checked in the native workbook exports. The importer reported unsupported style-extension elements; this does not establish complete formatting support. GUI entry, other locales and later-version execution were not tested.

SUM, SUMPRODUCT, COUNT.