A major modification package leaves for the authority with dozens of pages of justification and comes back with a single question: why was this test selected as the means of compliance rather than the calculation? The question is not a formality. It marks the exact point where the line runs between what an approved design organisation approves on its own and what must go up to the authority. That line depends neither on the manufacturer’s size nor on a cordial relationship with the caseworker: it follows a risk logic, and it is written into the regulatory texts.
Delegating Moves the Finding, Not the Decision
Everyday vocabulary speaks of “delegation,” as if the authority were lending out part of its signature. The actual mechanism is more precise, and it rests on three distinct acts that are constantly conflated.
The first is to show compliance: producing the test, the calculation, the analysis, the summary note that establishes a given requirement is satisfied. This act has never belonged to the authority. It is industrial by nature, and it already was back when authorities took a much more direct part in certification flight tests.
The second is to find compliance: reading the demonstration produced by the design office and attesting that it actually satisfies the certification-basis point it claims to cover. This is the act that gets delegated, and only this one. It is entrusted to individuals specifically named within the organisation, whose competence and independence from production the authority has accepted.
The third is to approve the design: deciding that the set of findings covers the certification basis and that the product, modification, or repair can be approved. This act remains an authority decision, with two important nuances detailed below: the approved organisation itself approves what is classified as minor, and it may approve certain categories of major changes when its approval grants it that privilege.
The authority doesn’t just hand over its power to approve designs. It lets companies show and verify compliance, but the final decision stays with the authority. Think of it like a teacher letting students grade their own homework, but the teacher still checks the most important parts.
Three Acts, Three Signatures That Must Not Be Confused
Within an approved organisation, this separation is made concrete through different signatures on different documents. The engineer responsible for the domain produces the test report or the calculation note and takes technical ownership of it. The designated certifying staff signs the compliance finding, and cannot sign one for a demonstration they produced themselves: internal independence is not an organisational nicety, it is the condition that makes the finding credible without the authority having to redo it.
The airworthiness review function, distinct from the design office, then checks that the file is complete, that the certification basis retained is the correct one, that the declared means of compliance have actually been followed, and that nothing has been left open. It is this function that carries the declaration of compliance to the authority, or that issues the approval when the organisation holds that privilege.
A practitioner coming from operations will recognise the structure: it is identical to the one that separates, in maintenance, the person who performs the task from the person who pronounces it fit to fly. The review is not a courtesy re-read, it is an act of a different nature, carried out by someone with no stake in the outcome of the demonstration.
Classification: The Lock That Holds Everything Else
The entire mechanism turns on one word: is the modification minor or major? A modification is minor when it has no appreciable effect on mass, balance, structural strength, reliability, operational characteristics, or other characteristics affecting the airworthiness of the product. Everything else is major. The wording looks flexible; in practice, it is the single most disputed point in the relationship between an organisation and its authority.

An approved design organisation classifies its own modifications, and approves the minor ones itself. This is the most visible benefit of the approval: a large volume of detail changes are handled without ever leaving the house. But the classification itself is a monitored act. The authority audits a sample of it, and when it judges that a file was classified too low, it does not simply reclassify it: it looks at the procedure that produced that classification.
| Act | Minor modification | Major without privilege | Major under privilege |
|---|---|---|---|
| Classification | Organisation, per approved procedure | Organisation, confirmed by the authority | Organisation, per approved procedure |
| Certification basis | Established by the organisation, framework set by the authority | Set by the authority | Set by the authority |
| Compliance demonstration | Organisation | Organisation | Organisation |
| Compliance finding | Designated certifying staff | Designated certifying staff, plus points taken up by the authority | Designated certifying staff |
| Approval | Organisation | Authority | Organisation, within the limits of its approval |
| Special conditions, deviations, equivalent level of safety | Not applicable in principle | Authority alone | Authority alone |
The trap: the major that builds up in small steps. Three successive modifications to the same system, each without appreciable effect taken in isolation, can together cross the major threshold — this is the classic case of repeated local mass gain, or the progressive tightening of a control law. If the classification procedure only reasons on the day’s delta, the organisation approves, in perfectly good faith, an evolution that should have gone up. The discovery generally comes much later, during an audit or an in-service event, and it then covers the entire series.
The right reflex: handle classification as a file in its own right, with its own note and its own reference. Include the actual starting state — the original approved configuration, not the previous modification — and track the cumulative effect system by system. And raise any new subject early: the authority’s level of involvement is fixed at the point when the certification programme is set, not at the end of the campaign.
The Level of Involvement Is Calculated, Not Negotiated
For a major file, the authority does not pick the points it will take up by guesswork. It applies a risk-based approach, subject by subject, crossing essentially four inputs: the novelty of the technical solution relative to the state of the art, its complexity, its criticality — that is, the consequence of a failure or a demonstration error — and the organisation’s performance in that domain, measured through audit findings and the feedback from previous files.
The result is not a uniform percentage applied to the file, but a map: on a single modification, the authority may settle for a sample review on installation aspects while remaining directly present on a single critical point, for example the demonstration of a particular failure mode or a flight test of handling qualities in a little-explored envelope.
The authority doesn’t just randomly check parts of a modification. It looks at how new, complex, and risky the change is, and how well the company has handled similar changes before. This helps decide how much the authority needs to be involved.
For a new aircraft model, the authority might fully delegate the certification of minor changes to the wing’s paint scheme (low complexity, low risk) but closely oversee the certification of a new flight control system (high complexity, high criticality). On a runway length of 2,500 meters at a temperature of 35°C, the authority might only sample-check the installation of new lighting systems (low risk) but require direct involvement in the certification of a new braking system (high risk).
The fourth input, the organisation’s performance, is the one most readily forgotten even though it is the only one within its control. An organisation whose files come back clean sees the authority’s involvement decrease on the subjects it has mastered. But it never falls to zero, and it rises again instantly as soon as a new subject appears: track record does not carry over from one technical domain to another.
What the Authority Never Lets Go Of
Beneath the variable level of involvement, there is an invariant baseline. No design organisation approval, however long-standing and well-rated, moves these points.
- The certification basis. Deciding which certification specifications apply, at which amendment, and what additional requirements are added, is an act of authority. The organisation proposes, the authority sets it.
- Everything that departs from the standard. A special condition drafted for a novel design feature, a finding of equivalent level of safety, an accepted deviation for an unmet requirement: these three instruments create rule. They cannot be signed by the party subject to them.
- Issuance of the certificate. The type certificate, its airworthiness data sheet, and supplemental type certificates outside privilege are issued by the authority.
- Mandatory continued-airworthiness measures. An organisation identifies an unsafe condition, proposes the corrective action, produces the service bulletin — but making the action mandatory for every operator worldwide falls to the authority alone.
- The organisation’s own approval. The organisation exposition, its procedures, the scope of its terms of approval, and the list of its privileges are approved by the authority. That is the very subject of the arrangement.
The Privilege Is Revocable, and That Is the Other Half of the Arrangement
A design organisation approval is not a permanent title. It comes with continuous oversight: scheduled audits, reviews of internally approved files, examination of non-conformities reported from service. When this oversight identifies a systemic failure — a classification procedure that does not work, a certifying staff member signing outside their domain of competence, insufficient traceability of the means of compliance — the response is not always withdrawal of the approval.
It is most often a narrowing of scope: the privilege to approve certain categories is suspended, or the level of involvement rises again on files in progress. For a design office, the consequence is immediate and is paid for in programme schedule, because every finding taken back by the authority adds to the critical path of a programme whose entry-into-service date, meanwhile, has not moved.
This is the point design teams most underestimate. The quality of the airworthiness management system is not a peripheral administrative constraint: it is the parameter that sets how many times the authority will step into the next programme.
What to Check in Your Own Documentation
Three concrete checks, whatever position is held in the chain. First, the exact scope of the terms of approval recorded in the approval: it names product types and disciplines, and a modification falling outside that scope cannot be approved internally, even if it is clearly minor. Second, the named list of certifying staff and the domain attached to each name: a finding signed outside its domain is a void finding, generally discovered at the worst possible moment. Finally, the classification procedure itself, and in particular how it handles the accumulation of successive modifications.
Two grey areas remain, and it is better to know them than to discover them. The first is the handling of approved operational data — flight manual, limitations, mandatory maintenance data — which follow a partly separate path from the design one and involve other contacts within the authority. The second is the interface with the production organisation: an approved design that cannot be produced in accordance with its data is worthless, and responsibility for this agreement between design and production is settled upstream, not at the time of first delivery.
For anyone preparing a certifying-staff nomination, the stakes can be summed up simply: what the authority examines is not only technical competence in the discipline, it is the ability to say no to one’s own organisation. That is the quality being delegated, and it is the only one the authority cannot verify from a distance.
This text describes a general mechanism and does not replace any official source: the applicable regulation, the certification specifications in force, and the organisation’s approved documentation prevail over this article in the event of a discrepancy.
Questions fréquentes
Can an approved design organisation approve a major modification on its own?
Yes, but only if its approval expressly grants that privilege and within the limits of its recorded terms of approval. Outside that scope, approving a major remains an act of the authority, even if the organisation has produced every demonstration and signed every compliance finding. The privilege is also revocable and can be suspended following unfavourable oversight.
What is the difference between showing compliance and finding compliance?
Showing compliance means producing the test, calculation, or analysis that establishes a requirement is satisfied; this act has always been industrial. Finding compliance means attesting that this demonstration actually covers the targeted certification-basis point. Only the second act is delegated, and it is entrusted to specifically named certifying staff, independent of the person who produced the demonstration.
How does the authority set its level of involvement on a programme?
It proceeds by risk analysis, subject by subject, crossing the novelty of the technical solution, its complexity, its criticality for safety, and the organisation’s past performance in that domain. The result is a map of points taken up and points delegated, not a uniform percentage. This level is fixed when the certification programme is established, not at the end of the campaign.
What happens if a modification was wrongly classified as minor?
The authority reclassifies the file, which requires reconstructing a proper certification basis and obtaining major approval. But the heaviest consequence falls on the classification procedure itself, which will be examined, and on the other files handled under the same logic. The consequences are then measured across a series of modifications, not a single file.
Does the age of an approval permanently reduce the authority's involvement?
It reduces it in the technical domains where the organisation has demonstrated reliable practice, down to a floor that never reaches zero. As soon as a new subject appears, involvement rises again, because a favourable track record does not carry over from one discipline to another. A long-standing approval therefore does not exempt an organisation from direct oversight on a novel design feature.
Which acts does the authority never delegate under any circumstances?
Establishing the certification basis, special conditions, findings of equivalent level of safety and accepted deviations, issuing certificates, mandatory continued-airworthiness measures, and the approval of the organisation itself. These acts create or apply the rule: they cannot be signed by the party subject to them.

Leave a Reply
You must be logged in to post a comment.