DMAX in Collabora Office: Find Selected Maxima and Keep Tie or Missing-Data Checks Separate

• Introduced: the original Collabora release and launch date are not established by the reviewed sources.
• Availability: documented in Collabora Office 26.04 desktop help.
• Platform scope: online and mobile deployment support is not inferred from the desktop documentation.
• Checked: October 14, 2026. Function-specific documentation and independent controls; no native Collabora execution.

=DMAX(Database; DatabaseField; SearchCriteria)

All arguments are required. The dedicated reference excludes blank and nonnumeric field cells. It documents zero when no records match or no nonzero numeric field values remain.

A maximum value does not itself identify the record that produced it, count tied records or prove that valid measurements were present. Those are separate reporting questions.

Enter this database in A1:E8, including the header row. Cells marked leave empty must be physically empty; do not type those words or enter an empty-string formula. Enter pending as ordinary text.

Put the exact header Cost in G1 and the criterion >=10 in G2. The criteria range is G1:G2, including its header. It selects items A, B, F and G. Its column need not be vertically aligned with the database’s Cost column.

Use an empty output cell outside the database and criteria areas for each formula. Do not include unused rows in these criteria ranges. The database range includes all seven records, even though only four satisfy this first criterion.

With Cost >=10, compare:

=DMAX(A1:E8;"Units";G1:G2)
=DMAX(A1:E8;"Cost";G1:G2)

The first result is 6, from item F. The second is 20, from item B. B’s recorded Units value is zero, so the record with the maximum Cost does not also have the maximum Units.

Keep the measured field in the report label. A bare “maximum” can conceal which quantity was compared.

Put Cost in G4 and H4, >=10 in G5, and <16 in H5. The repeated-header AND criteria select A, F and G.

=DMAX(A1:E8;"Cost";G4:H5) gives 15. Maximizing Units for the same records still gives 6. The excluded B record had the highest cost but not the highest quantity.

The header repetition defines a bounded interval on Cost. Moving those two comparisons to separate rows would change the rule to OR.

In a separate copy, change A’s Units from four to 6. With the original Cost >=10 criteria, the maximum is still six, but A and F now tie.

DMAX returns a value rather than a unique item. It cannot establish that F is the only maximum-producing record merely because the unmodified example had that property.

The numeric Units count remains 3, and the matching record count remains 4. Restore A to four before checking the remaining original-table outputs.

Use a separate two-column database in A12:B15, with headers Key and Adjustment and records X with numeric input -8, Y with -3, and Z with -5. Put Key in D12 and <> in D13 to match nonempty keys.

=DMAX(A12:B15;"Adjustment";D12:D13) has expected −3. It is the largest of the three negative observations. A maximum does not add an artificial zero candidate when valid negative numbers exist.

DMIN on the same database gives −8. The numeric span is 5, calculated as −3−(−8), not by subtracting magnitudes in the opposite order.

Change only Y’s Adjustment in that separate database to zero. DMAX now gives 0, and DCOUNT on Adjustment gives 3: zero was actually observed in a numeric set.

A valid criterion selecting no records also produces the documented DMAX zero, but DCOUNT gives zero. Keep a numeric count beside the maximum when a missing-data state must not masquerade as a measured zero.

The same recorded maximum can therefore have different meanings depending on whether numeric observations were selected.

A wrong field name or invalid field index is an argument problem documented in the shared help. It is not a legitimate zero maximum. Use exact headers and inspect criteria ranges before adding any error fallback.

Dates and logical values have the database category’s explicit numeric treatment; this guide uses ordinary numeric adjustments and quantities instead. Formula-empty text, competing errors and a complete conversion matrix remain unverified. The examples are independently calculated, with no native application execution.

The field argument selects the column to calculate after records are matched; it need not be the column used in the criteria. It can be a quoted header, a reference to its header cell, or a one-based column index within the database.

For this table, Units, C1 and index 3 identify the same field. Index three means the third database column, not an absolute worksheet-column number. The shared help says a numeric field index ignores digits after the decimal point, but the examples use integers for clarity.

Criteria in the same row are combined with AND; criteria in separate rows are combined with OR. Repeating a header lets two conditions apply to the same field. Text matching and wildcard or regular-expression behavior depend on Calc’s calculation settings. The primary numeric criterion avoids those pattern choices.

If you use literal Team criteria, enable whole-cell matching and use the exact labels shown. Do not silently treat wildcard and regular-expression modes as interchangeable. The shared database documentation also treats dates and logical values as numeric in these functions; that rule should not be replaced with another aggregation function’s type rules.

• Collabora Office 26.04: DMAX.
• Shared database arguments and criteria rules.
• Editorial testing standards.

Complete function-specific and shared database documentation reviewed. Outputs are independent row-selection and arithmetic controls or explicit documented rules, not captured native results. Formulas use English names and semicolon separators.