Peer review

Caravel Press operates double anonymous peer review. Reviewers are not told who the authors are, and authors are not told who the reviewers are.

This is not a setting that can be switched off. Author identity is held in a separate part of the system from the manuscript, and the reviewer's copy is assembled without it, so a reviewer cannot see an author's name, affiliation or funding even by accident.

1. 1. Checks

   The editorial office confirms the submission is complete, in scope, and correctly anonymised, and screens it for similarity with iThenticate. An incomplete submission is returned rather than rejected.

2. 2. Editorial assessment

   A handling editor reads the paper and decides whether it should go to review. A paper outside the journal's scope, or clearly below the standard the journal publishes, is declined at this stage with reasons. This is a decision on scope and standard, not a shortcut around review.

3. 3. Review

   At least two independent reviewers with subject expertise assess the work. Reviewers are selected by the handling editor for expertise and independence: nobody at the authors' institution, no recent collaborator, and nobody with a declared competing interest. Authors may suggest reviewers and may ask that particular people are not approached; the editor is not bound by either.

4. 4. Decision

   The handling editor makes a recommendation on the reviews. The final decision rests with the editorial office and is one of: accept, minor revisions, major revisions, revise and resubmit for further review, or decline. Reviewers advise; editors decide.

5. 5. Revision

   Authors receive the full text of every review and respond point by point. A revised paper normally returns to the same reviewers. There is no limit on the number of rounds, but a paper that is not converging will be declined rather than kept in circulation.

A paper can be withdrawn at any point before it is published. Write to [email protected] from the address it was submitted under and say which submission you mean. No reason has to be given.

• Nothing is charged for a withdrawal, at any stage, including after acceptance.
• The paper is taken out of the review process, and any reviewer still working on it is told it has been withdrawn.
• The submission and its history stay on the editorial record. They are not public and they are not shared.
• Withdrawal is what frees a paper to go elsewhere. Until it is withdrawn it is still under consideration here, and submitting it to another journal in the meantime breaks the undertaking given at submission.

Once an article is published it cannot be withdrawn. The published record is permanent and the DOI goes on resolving. The only instruments after publication are correction, expression of concern and retraction, set out in the corrections and retractions policy.

Every review scores the paper from 1 to 5 on originality, methodology, significance, clarity and relevance, and requires a written comment on each. A score without a reason is a verdict with no reasoning, and the system will not accept one.

Reviewers also give overall comments to the authors, and may add a confidential note to the editor which the authors never see.

To respond to an invitation

4 working days

To complete a review

14 calendar days

Submission to first decision

Measured and stated here once enough papers have completed review

Submission to publication

3–5 weeks, the median across 314 papers published by Caravel Press

Every published article carries the date it was submitted, the date it was accepted and the date it was published.

The last two figures above are measured rather than undertaken, and then given room. Submission to publication is the median across the papers the journal has published, recomputed every time this page is built, with a week of margin added at each of the three stages. A median is the middle of the record, so half of what the journal publishes takes longer than it, and a figure an author plans around should sit above the middle rather than in it. Where an article's submission and publication dates fall on the same day it was carried across from the system this one replaced, with no submission date of its own, so it is treated as unmeasured rather than as having taken no time at all: counting it as nought would drag the median below anything a paper has actually taken. Where an author asks for their article to wait for a later volume, the days it waits at their request are left out of its time to publication, and nothing else is.

A journal's own page also breaks the time down by stage. Those parts are measured in the same way once the decisions behind them have been recorded here. Until a stage has enough of them it shows the editors' estimate of how the measured time divides, labelled as an estimate beside the total, and it is replaced by its own median as soon as there is one. The same week of margin is added to every stage, measured or estimated, so the parts always add up to the figure beside them.

A special issue is reviewed to exactly the same standard, by the same process, in the same system as any other submission. There is no separate route and no lighter touch.

• Guest editors are appointed by the editor-in-chief and are named on the collection.
• A guest editor cannot handle a paper they have authored or co-authored, and cannot see it in the system.
• Final decisions on every paper in a special issue rest with the journal's own editors, not with the guest editors.
• A collection is not a route for its guest editors to publish their own work at volume, and the proportion of it they author is watched as closely as it is across the journal.