Nobody Owns the Corrections Window

The viva went well. The examiners recommended the award subject to corrections, everyone shook hands, and the candidate walked out with a list.

Now consider what happened to that person in the same week. The stipend ended at submission. The formal supervision entitlement is tied to a registration status they no longer hold. The writing centre books by course code and they are no longer on one. They have moved city for a job that starts on the strength of a degree that has not been conferred. And they have somewhere between four weeks and a year — if their institution says at all — to rewrite a document to a standard set by two people they will not meet again.

Your graduate school carries the completion statistic for that period. No service in your institution is tasked with it.

The window is real, bounded and inconsistently published

Where institutions state a deadline, they state it precisely, and the spread is wide enough that a sector average would be meaningless. Every figure below was read first-hand from the institution’s own page on 24 August 2026:

Three details in that table are worth pausing on. Bristol’s minor-errors deadline is 28 days, not a month of working time — and it runs from an email rather than from any point the candidate controls. Bristol is also explicit about scope: when making corrections, a candidate must only address the points raised in the feedback. That is sound regulation, and it also means the candidate cannot simply rewrite their way out of an unclear instruction.

Sheffield adds the detail most likely to catch an institution out. Its windows apply regardless of whether they are registered as full-time or part-time. A part-time candidate — often the one holding down a job precisely because the funding ran out — absorbs identical rework into a fraction of the available hours.

And the Edinburgh rows are the most revealing of the three, because of what is not there. Its guidance states that The deadline for your corrections depends on the recommendation that was made by your examiners, and the only three-month figure on the page attaches to minor corrections following a resubmission. A candidate reading it the week after a first viva learns no date at all. That is not a criticism of one institution — it is the clearest available evidence for this article’s argument. A stage nobody owns is also a stage nobody is required to put a number on.

The second cost is on the examiner side and is almost never counted. Edinburgh states that Examiners are permitted 4 weeks to review the corrections, but this will depend on examiner availability. That is a second read of a long document by senior academics, scheduled against nobody’s workload model.

One caution on using any of this. We quote only institutions whose own regulations we could open and read on the day, and we have marked the gap where a figure does not exist rather than filling it from a summary. Correction deadlines vary widely and many sit in documents that are not publicly readable. A number repeated from a search result rather than from the regulation is exactly the kind of figure that ends up wrong in a committee paper. Read your own regulations.

Three entitlements that end at the viva

The reason the window is unowned is not neglect. It is that three separate systems all key off a status that changes on the day of the examination.

1. Funding. Stipends and most fee waivers end at submission, not at conferral. The candidate is doing mandated academic work with no income attached to it.
2. Supervision. The formal obligation is normally expressed as a duty to a registered candidate. Most supervisors keep helping anyway, out of decency rather than workload allocation — which means the support is real, unrecorded, and unevenly distributed depending on who your supervisor is.
3. Services. Writing centres, disability support and skills programmes are provisioned against enrolled cohorts. A candidate in corrections is administratively adjacent to all of them.

That pattern is the same one that produces the earlier gap described in why your longest-registered candidates are the ones nobody is supporting. The corrections window is that gap again, shorter and with a hard deadline on the end of it.

It is also, formally, part of the assessment. The QAA guidance on assessment lists among its principles that Students are supported and prepared for assessment and that Assessment is timely. Corrections are a stage of the examination, not an administrative postscript, and the support principle does not stop applying because the viva is over.

The part nobody measures: what the corrections were actually about

Ask a graduate school how many candidates received minor corrections last year and you will get a number within the hour. Ask what those corrections were about and almost nobody can answer, because almost nobody codes the lists.

This matters more than it sounds, because the four things a correction can concern have completely different owners:

• Apparatus — inconsistent references, a bibliography assembled by hand over four years, citation style drift between chapters.
• Structure — a contents list that no longer matches the pagination, a missing figure list, sections in a non-conforming order.
• Prose — passages an examiner could not follow, or claims phrased more strongly than the results section supports.
• Argument — the interpretation does not follow from the findings, or a limitation was not addressed.

Only the last of those is a research problem. The first two are properties of how the manuscript was authored and were fixable three years earlier. The third sits between: some of it is genuine analytical work, and some of it is the claim-strength question that two examiners will read differently — which is precisely the marker-disagreement cause set out in how to get consistent marking on extended written work.

Until that split is measured, every conversation about corrections is a conversation about anecdotes. And the split is genuinely cheap to produce: it is one afternoon with the correction lists from your last two conferral rounds and a four-column tally sheet.

Why the fix cannot live inside the window

The instinct is to staff the window: a corrections clinic, a dedicated adviser, an extension policy. Each helps a little and none of them addresses the cause, for the same reason a longer deposit checklist does not fix deposit returns.

An apparatus correction exists because the references were assembled manually. A structure correction exists because the headings were never styled. Both were determined years before the viva, and both are discovered at the point in the process with the least support attached and the hardest deadline. That is the same upstream pattern documented in why your deposit desk is holding degrees for metadata, arriving one stage earlier.

And be careful about which tools you expect to help. A grammar checker or a paraphraser works on the sentence in front of the cursor, which means it can reach the intelligibility half of a prose correction and none of the apparatus, structure or argument categories — the level-by-level version of that argument is in what academic writing tools actually check.

Three counts you already hold

Before any purchase, three numbers from the last two conferral rounds. All three come from records your examinations office already keeps.

1. Outcome distribution. How many candidates received no corrections, minor corrections, major corrections and resubmission.
2. Elapsed time against the permitted window. Median days from decision to approved corrections, expressed as a share of the limit. A median above 80% of the window means your candidates are finishing against the wall, and any variation will produce misses. If your regulations state no limit, that is your first finding.
3. The coded split. Correction items tallied into apparatus, structure, prose and argument.

Number three is the decisive one. If argument dominates, this is a supervision and assessment-design question and no platform will help you — say so and spend the budget on supervisor development. If apparatus and structure dominate, you have an authoring problem wearing an examination costume, and it is solvable.

What Tesify for Institutions removes from the window

The premise is the same one that applies at deposit: satisfy the mechanical requirements while the thesis is being written, so the correction list that arrives is about the research.

• A consistent reference apparatus maintained as the manuscript grows, in a named published style, rather than reconciled across four years of accumulated entries in the final month.
• Generated front matter — contents, figure and table lists produced from the document’s own structure, so they cannot drift out of step with pagination.
• Section order held against your published specification from the first chapter, not audited at the end.
• Structured source, which is what makes an accessible archival export possible and removes a whole class of late-stage repair.
• A legible drafting history, so a supervisor reviewing chapter four is looking at progress rather than only at outcomes — the workload pattern described in what your supervisors are actually spending review time on.

None of that touches the scholarship, and it should not. It removes the corrections that were never about the research from a window in which the candidate has no funding, no formal supervisor and no service to call.

We would not claim it reduces argument corrections. Those are the ones you want examiners to be making.

Start with one department and one measure

The next step is a free departmental pilot in one graduate programme for one conferral cycle, with the coded correction split agreed as the single measure before it begins. That is a comparison against your own prior rounds, in your own regulations, and it needs no tender, no capital request and no institution-wide commitment — which is exactly why it clears governance when a purchase order does not.

If your coded split comes back dominated by argument, we will tell you that a platform is not your answer. That is a better outcome than a licence nobody can evidence.

Request an institutional evaluation and we will work through your outcome distribution, your elapsed-time median and a single-department pilot with your graduate school team.

Frequently asked questions

How long do candidates get to complete thesis corrections?

It varies by institution and by outcome. Bristol gives 28 days for minor errors, usually six months for errors of substance and usually twelve months for a resubmission. Sheffield gives three months for minor corrections, six months for major corrections and one year for a full resubmission. Edinburgh publishes no single figure for a first examination, stating instead that the deadline depends on the recommendation the examiners made. Always read your own regulations rather than a sector average.

Why is the corrections window described as unowned?

Because the three things that carried the candidate through writing up typically end at the viva: funding, the formal supervision entitlement tied to registration status, and routine access to student services. The stage still has a deadline and a workload, but no service is tasked with it.

Do corrections count as part of assessment?

Yes. Corrections are a stage of the examination, not an administrative afterthought, which is why the QAA principle that students are supported and prepared for assessment applies to them as much as to anything earlier in the process.

Does every institution publish a corrections deadline?

No, and that is itself worth noting. Edinburgh’s College of Arts, Humanities and Social Sciences states that the deadline for corrections depends on the recommendation the examiners made, and gives a three-month figure only for minor corrections following a resubmission. A candidate reading that page after a first viva learns no date at all.

What share of corrections are about writing rather than research?

Almost no institution knows, because almost none codes its correction lists. The categories that would answer it are apparatus, structure, prose and argument, and separating them is a one-afternoon exercise on the last two conferral rounds.

Does the corrections window cost examiner time as well?

Yes, and it is rarely counted. Edinburgh states that examiners are permitted four weeks to review the corrections, but that this will depend on examiner availability. That is a second read of a long document by senior academics, scheduled outside any workload model.

Can a candidate go beyond what the examiners asked for?

No, and the regulations are usually explicit. Bristol states that when making corrections you must only address the points raised in the feedback. A candidate who rewrites more than was asked creates a second review problem for the examiners.

Does the deadline change if a candidate is part-time?

Not necessarily. Sheffield states that its three-month minor and six-month major correction windows apply regardless of whether the student is registered full-time or part-time, so a part-time candidate absorbs the same rework into far less available time.

Why do corrections delay conferral so often?

Because the corrections deadline, the examiner re-read and the deposit check all sit inside fixed calendar deadlines with no slack. A correction completed late in its window can miss the conferral round entirely rather than delay it by the same number of days.

Which candidates are most exposed?

Those whose funding ended earliest, those who have already moved city or country for work, and those writing in an acquired language who receive a larger share of prose-level corrections. The consequences of a moved conferral date also fall hardest on candidates whose visa or job offer depends on it.

What three numbers should we pull first?

The distribution of examination outcomes over the last two rounds, the median elapsed time from viva to approved corrections against the permitted window, and the correction items coded into apparatus, structure, prose and argument. All three come from records you already hold.

Is this a supervision problem or an authoring problem?

The coding exercise decides it. If argument dominates, it is a supervision and assessment-design question and no platform helps. If apparatus and structure dominate, it is an authoring problem that was fixable years before the viva.

What would a pilot on this look like?

One conferral cycle in one programme, with the coded correction split agreed as the measure before it starts, compared against your own prior rounds. It is scoped to a single department so that it needs no tender and no capital request.