# 6.13.2 Two parameter sets, and they do not mix

The clause gives a parameter table for the two named sets, Ed25519 and Ed448. Each row fixes the field, the encoding length, the cofactor, the octet length, the two hash functions, the two curve coefficients and the base point. The larger set also defines a domain separation string that carries a caller supplied context of up to 255 octets.

## What it asks

Pin one parameter set per deployment, Ed25519 or Ed448, and record every value from the table, including the base point and its order. Where the larger set is used with a context, define how the context is built and bounded, and test the boundary at the maximum length. Never let one deployment mix values from the two sets.

## What the amendment does

Inserts 6.13.2 and its table, after 6.13.1.

## Who it binds

The deployment owner, the configuration manager and the test owner.

## Evidence an auditor samples

- The parameter record, one row per deployment, every table value present (configuration)
- The context construction rule and its length bound (interface specification)
- The test vectors for both sets, run in continuous integration (test suite)

## The test you can run

Ask a deployment which set it uses and for the hash pair. If the answer mixes a value from the other set, the deployment is not one of the two specified mechanisms.

## Common failures

- The deployment is called Ed25519 while a parameter comes from the other set.
- The context string is built without a length bound.

## How to close it

Copy the table into the configuration record rather than referencing it. A reviewer who has to open a purchased standard to check a deployment will not check it, and the record is what the auditor samples.

## Source

ISO/IEC 14888-3:2018/FDAM 1:2026(en), draft, 6.13.2. Read on 21 October 2026. Our wording, not the instrument's.
https://www.iso.org/standard/89517.html

Not legal advice; does not certify.