# Surface roughness Rz

Interpreting profile peak-to-valley roughness with explicit ISO or legacy definitions, filtering and measurement conditions.

Reviewed: 2026-11-01

Tags: parts-and-design, specifications, surface-texture, metrology

## Parameter identity and standard version

Rz is a profile height parameter. Identify the standard and edition, filter and evaluated length before interpreting the symbol.

In the idealized averaging example, five filtered section heights of 4, 5, 6, 5 and 10 µm give **6 µm**. The largest section height is 10 µm. Overall Rt requires the highest peak and deepest pit over the full evaluation length, which these five ranges cannot supply. The arithmetic assumes an operator that averages the section heights.

The current published reference checked here is **ISO 21920-2:2021**, edition 1, December 2021, **corrected version June 2022**, for profile terminology and parameters. ISO lists ISO 4287:1997 and its listed amendment/corrections as withdrawn predecessors. **ISO 21920-3:2021**, edition 1, December 2021, addresses specification operators and replaces ISO 4288:1996. Both current records indicate revision work; their successor committee drafts are not published replacements. [ISO Part 2](https://www.iso.org/standard/72226.html), [ISO Part 3](https://www.iso.org/standard/72228.html)

Withdrawal does not authorize silently reinterpreting an existing drawing. Digital Surf explains that older drawing specifications remain tied to their former standards, and that changing the evaluation workflow can change results. Establish the intended specification before updating an inspection program or copying a roughness limit into a new drawing. [Digital Surf comparison of ISO 4287 and ISO 21920](https://www.digitalsurf.com/blog/what-are-the-differences-between-iso-4287-and-iso-21920/)

## What the height represents

Mitutoyo's ISO 21920 parameter guide describes the section-height contribution as the highest profile peak plus the deepest profile pit within a section length. Rz retains section-based evaluation and averaging, unlike parameters now evaluated directly over the evaluation length. Keep the section length and aggregation rule explicit. [Mitutoyo guide, printed page 53](https://www.mitutoyo.com/webfoo/wp-content/uploads/E11003-Quick-Guide.pdf), [Digital Surf section-length explanation](https://guide.digitalsurf.com/en/guide-iso-21920-parameters.html)

| Quantity or convention | Distinction to preserve |
| --- | --- |
| Section peak-to-valley height | Highest peak plus deepest pit in the specified section |
| Rz result | Section-based parameter evaluated under the selected standard/operator |
| Rt | Total peak-to-valley height over the evaluation length, not the same averaging operation |
| Old JIS/ISO ten-point Rz | A historical parameter, not interchangeable with the current maximum-height convention |

Mitutoyo explicitly warns that old JIS and ISO 4287-1:1984 used Rz for ten-point height and that the differences are not necessarily negligible. Taylor Hobson describes that legacy ten-point calculation using the five highest peaks and five deepest valleys within the sampling length. Resolve labels such as “Rz JIS,” “Rz DIN” or an instrument's legacy mode before comparing certificates. [Mitutoyo, printed pages 52–53](https://www.mitutoyo.com/webfoo/wp-content/uploads/E11003-Quick-Guide.pdf), [Taylor Hobson amplitude parameters, July 2022 article](https://www.taylor-hobson.com/resource-center/blog/2022/july/what-are-surface-finish-amplitude-parameters)

Do not relabel a single scratch depth, a whole-part height range or an areal measurement as Rz merely because the units are micrometres. Identify the profile and calculation actually used.

## Filtering, lengths and a bounded settings example

Roughness is evaluated from a processed profile. Renishaw describes removal of underlying geometry and separation of roughness from waviness by filtering. Therefore an unchanged physical surface can produce different reported parameters when the profile-processing settings change. [Renishaw surface-finish measurement](https://www.renishaw.com/cmmsupport/knowledgebase/en/surface-finish-measurement--22135)

As a **specific ISO 21920-3 default-table example reproduced by Mitutoyo**, an Rz upper tolerance limit of **10 µm** lies in setting class **Sc3**. That column specifies an L-filter nesting index of **0.8 mm**, evaluation length **4 mm**, S-filter nesting index **2.5 µm**, maximum sampling distance **0.5 µm**, maximum nominal tip radius **2 µm**, and five **0.8 mm** sections. This is a bounded table example, not a universal setting for every Rz requirement. Explicit drawing settings and the applicable operator must be checked. [Mitutoyo guide, printed page 55](https://www.mitutoyo.com/webfoo/wp-content/uploads/E11003-Quick-Guide.pdf)

Record the actual filter type and nesting indices/cutoffs, section and evaluation lengths, form-removal operation and software mode. Do not shorten a trace to fit a small land and report the result as though the original operator remained unchanged. Resolve restricted access or insufficient length with the specification owner.

## Direction and instrument effects

Measurement direction matters on a surface with lay. KEYENCE's procedure, explicitly presented as an excerpt from **ISO 4288:1996**, calls for cleaning and, where direction is unspecified, choosing the direction giving the largest height parameters. Use this as legacy procedural guidance; it does not independently establish every current ISO 21920 default. Preserve the specified direction and trace location when reproducing a result. [KEYENCE stylus measurement procedure](https://www.keyence.com/ss/products/microscope/roughness/line/measurement_procedure.jsp)

The stylus also mechanically filters the surface. Taylor Hobson explains that a finite-radius tip can bridge narrow depressions, reducing its response to some features. Tip geometry is therefore part of the measurement definition, not merely an interchangeable accessory. [Taylor Hobson Form Talysurf Intra handbook, printed page 120](https://www.taylor-hobson.com/-/media/ametektaylorhobson/files/learning-zone/user-guides/form-talysurf-intra-handbook-1_5.pdf)

Check the instrument's calibration, tip condition, tracking and usable measurement range. Keep contact-force and optical-response differences in view when comparing methods. Agreement of displayed units does not establish agreement of spatial bandwidth or feature accessibility.

## Original bounded calculation

Assume five equal sections of an already correctly filtered profile, using an operator that averages their peak-to-valley heights. Let those heights be **4, 5, 6, 5 and 10 µm**:

`mean section height = (4 + 5 + 6 + 5 + 10) / 5 = 6 µm`.

The largest section height is **10 µm**, so it is not the same quantity as that **6 µm** average. Nor can Rt be reconstructed from these five numbers alone: the highest peak and deepest pit over the full evaluation length may occur in different sections, and the table supplies only their within-section sums.

For example, sections with peak/pit magnitudes `(9,1)` and `(1,9) µm` each have height **10 µm**, yet together contain an overall peak-to-valley height of **18 µm**. These are explanatory profile values, not raw instrument data or acceptance limits.

## Ra is not a fixed conversion

Ra summarizes absolute profile ordinates; Rz responds to section extremes. Mitutoyo presents different definitions for the two. A single average cannot reconstruct the distribution of peaks and pits, so a fixed Ra-to-Rz multiplier is not an identity. A process-specific empirical relationship may assist planning only within its demonstrated scope; it should not replace measurement against the specified parameter. [Mitutoyo parameter definitions, printed pages 52–53](https://www.mitutoyo.com/webfoo/wp-content/uploads/E11003-Quick-Guide.pdf)

For a useful handoff, retain the part/revision, parameter and standard edition, numerical limit and units, operator settings, trace direction/location, instrument/tip, evaluation software and decision rule. Keep the measured trace where needed to explain an outlier or a disagreement between methods.

## Source versions

Mitutoyo's downloaded guide prints **Catalog E11003(7)** and **2408 (JP12)** on its back cover; no calendar issue date is inferred here from that code. The Taylor Hobson handbook prints **Issue 1.5, June 2002**. Its physical stylus explanation is used without treating its legacy parameter tables as current ISO defaults. Live tutorial copyright years are not asserted as technical revision dates. The arithmetic examples are original.

[Dimensions, tolerances and specifications](index.md)