A reviewer reading our vug-quantification paper wanted one thing the manuscript did not give them: the fluid numbers. To measure individual vugs on a borehole-image log, the detector keys on the resistivity contrast between a pore and the rock around it, and that contrast is set by physics the paper never stated outright, namely how salty the formation water is, how conductive the invaded mud filtrate is, and how resistive the mud cake on the borehole wall is. Without those, a careful reviewer cannot check whether the contrast the method relies on is real or an artefact of processing. So they asked for the connate-water salinity and the mud conductivity, and the request was fair.
The problem is that those numbers are real measurements from a producing field, and the field belongs to a mid-sized Middle East carbonate operator we partnered with under a confidentiality agreement. Printing formation-water salinity and mud resistivities in a petroleum-engineering journal would put an operator's fluid measurements on the permanent public record, and the agreement forbids exactly that. So we had a reviewer who needed the physics to sign off, and a contract that would not let the physics onto the page. This piece is about the maneuver that resolved it, and where the line between the two sits.
The two channels a paper travels through
The mistake is to treat "the paper" as a single object with one confidentiality setting. It is not. A manuscript under review moves through two separate channels, and they have different rules.
The first is the manuscript itself: the PDF that, if accepted, becomes a citable public artefact indexed and downloadable forever. Anything printed here is disclosed to the world.
The second is the review correspondence: the letters between authors and reviewers, mediated by the editor, that never get published. Reviewers are bound by the journal's confidentiality norms; the correspondence is not a public document. What travels down this channel reaches the handful of people qualified to judge the work and stops there.
Those two channels have different confidentiality boundaries, and the whole maneuver lives in that gap. A physical parameter a reviewer needs to verify the method can ride the private channel to the people who need it, while the manuscript carries none of it. The reviewer's scientific need and the operator's confidentiality are only in conflict if you insist that every answer must appear in print. Answer in the correspondence instead, and they stop conflicting.
What we handed over, and what we did not
The payload was three numbers, disclosed to the reviewers and kept out of the manuscript. The formation-water salinity, , which sets how conductive the connate water in the pores is. The mud-filtrate resistivity, , the invaded fluid's conductivity near the wall. And the mud-cake resistivity, , the conductivity of the layer the imaging tool presses against. Together those three fix the resistivity contrast a vug shows against its matrix on a dynamic image, which is precisely what the reviewer needed to sanity-check the detector's premise.
What we did not hand over is the line that matters. Not the field images, not the raw well data, not a depth digit. The three parameters are physical constants of the fluid system; they describe the physics without identifying the field or exposing the corpus the operator guards. Disclosing them privately answers the science question while leaving the confidential asset untouched. That is the distinction the whole exchange turns on: a fluid property is not the field data, and a reviewer who needs the former does not need the latter.
The wall is the NDA storage clause
What makes the private channel legitimate, rather than a convenient dodge, is that the confidentiality agreement is explicit about where confidential information may live. The clause that governs the whole engagement requires that confidential information be stored only on local computers and devices, not on a remote server of any kind, reachable on a local network but never internet-accessible, password-protected, with hardware in locked cabinets when not in use. It is a storage rule, and it is doing the load-bearing work here.
Read that clause against the two channels and the boundary draws itself. The field data has to sit on local, non-internet-accessible, password-protected machines and cannot leave that perimeter, which means it cannot go into a manuscript, an email attachment to a reviewer, or a public repository. A salinity value is not the field data; it is a derived measurement of the fluid, and disclosing it to a reviewer in correspondence moves the guarded corpus nowhere. The storage clause is the wall in the exhibit above: it is why the parameters can cross into the review channel while the field data crosses into no channel at all. The rule that keeps the corpus on locked local hardware is the same rule that authorises the private disclosure of a number derived from it.
Where the line sits, precisely
The clean version of the boundary is worth stating in one place, because it is the transferable part.
A physical parameter that verifies a stated method can be disclosed to reviewers in the correspondence and withheld from the manuscript. The three fluid numbers are exactly this. They let a reviewer confirm the resistivity contrast is genuine, and they never appear in print.
Field data itself crosses no wall. The images, the raw logs, the depths that would identify the acreage stay on local devices under the storage clause, out of both the manuscript and the correspondence. A reviewer who asked for the raw well data rather than a fluid property would have hit a request the contract does not permit.
The test we used for the judgement call between those two is whether a value describes the physics or identifies the asset. A salinity describes the physics. A depth, a well name, a field image identifies the asset. When a request sat on the physics side, the review channel carried it; when it touched the identity of the field, it stopped at the wall.
This is one request inside a larger reviewer exchange. The full concession map for that second revision, which comments we defended, conceded, or answered off the page, is worked through in What Four Reviewer Challenges Actually Asked Us; the broader strategy of publishing peer-reviewed work at all under a national-oil-company agreement is in Publishing Peer-Reviewed AI Research Under a National-Oil-Company NDA. This note isolates one maneuver because it generalises past our field: any team publishing on confidential industrial data will eventually meet a reviewer who needs a number the client will not let them print, and the private-channel disclosure keeps both the science and the contract intact.
The paper went to a vote and its fracture-and-bedding sibling from the same programme was accepted at a petroleum-engineering journal and entered copyediting. The reviewer got a complete answer to the physics question, and the operator's field data never left the locked local machines it is required to sit on, because the answer went into the correspondence and not the paper.
Limitations
This is one disclosure inside one paper's revision, not a general procedure for reviewer response. The move worked because the requested values were fluid properties that describe the physics without identifying the field; a request for the raw well data, the field name, or anything that pins the acreage would have had no analogous path and would simply have been declined under the agreement. The legitimacy of the private channel rests on the specific storage clause in this engagement, which keeps confidential information on local, non-internet-accessible, password-protected devices; a differently drafted agreement, or a journal with different correspondence-confidentiality norms, could shift where the wall sits. And the account is anonymised throughout: the parameter values and the storage-clause wording are reported, while the operator, the field, and the wells are held back under the agreement that shaped the exchange.
References
[1] Committee on Publication Ethics (COPE), guidelines on peer-review confidentiality. https://publicationethics.org/
