Anyone who wants to say how long an orphaned well's leak will last, or how much gas sits behind it, reaches first for the well's production history. Fit a decline curve to the last years of gas sales, extend it forward, and the answer looks like engineering. For producing wells it is the standard tool, and for good reason: a well that reported its gas every month for twenty years has a record that says a great deal about its reservoir.
For orphaned wells we went looking for that record across Oklahoma's public files. Mostly it is not there; the routes we tried to rebuild it did not work; and where a history does survive, it answers a different question from the one a leak asks.
Most orphaned wells were drilled before the record began
The Oklahoma Corporation Commission keeps two lists of wells with no responsible operator: the orphan list and the State Funds plugging list. In September 2026 they held 21,697 wells between them, and we checked every one against the public completion and production records.
83%of the 21,697 wells on Oklahoma's orphan and State Funds lists were drilled or first produced before July 1990
July 1990 matters because that is when the Oklahoma Tax Commission's electronic production record begins. Whatever a well sold before then survives, if at all, on paper.
The second problem is the unit of reporting. Oklahoma records production by lease, not by well. A lease may carry one well or a dozen, and the state keeps one monthly volume for the lease. Fewer than one in five of the listed wells has any lease production identifier on record, and about one in ten sits on a lease carrying only that well. That one in ten is the ceiling on how many could ever have a state production series that is theirs alone, before anyone asks whether the series is long or complete enough to fit.
We tried to rebuild it two ways
The first route was commercial production data keyed to each well's own identifier. The keys matched cleanly. However, the data had been assembled to find gas worth producing, so it leaned heavily toward large producers and covered a different part of the state from the wells we work on. Across the roughly 22,300 Oklahoma wells in our database, it would have given a first gas figure to 153.
The second route was the state's own bulk production files, which are keyed by lease. Giving a lease's volume to one of its wells means matching names and locations between two record systems that were never designed to meet. When we checked the matches, the assignment named the wrong unit about one time in four. Of the 15,770 wells in our database with no gas figure, 4 gained a well-level history we would stand on.
The second failure is the more useful one to understand. A lease volume attached to a single well looks like a measurement, with dates, units and decimals, but it is not about the thing it is attached to, and nothing in the number warns you.
The same trap caught our own data
In 2025 we collected a set of per-well production charts. They had been matched to wells by the order of their file names rather than by the well each chart described, so each history sat under another well. Nothing looked wrong: the volumes were plausible, and so were the declines.
We found it this year by cross-checking against a second, independently keyed source. Of the 73 charts with an exact counterpart in that source, none sat under the right well. We dropped every series the second source could not corroborate, re-ran the analysis that had used them, and corrected the figures we had already shared. The number of our measured wells with a usable history of their own fell by about half.
What we took from it is specific: a production history for an orphaned well is a claim about identity before it is a claim about gas, and it needs a second, independent record before anyone leans on it.
Where a history exists, it describes how production ended
Some orphaned wells do have a series of their own. The ones we have examined rarely show the smooth decline a curve expects. They have years of zeros while the well sat shut in, revivals after a workover or a recompletion, rises late in life, and endings where a steady rate went to nothing within a month or two.
A decline fit on a record like that is unstable. Fit the curve to one stretch and it falls; move the stretch a few years either way and, on most of the series we could fit, the fitted trend changed sign. No rule in the record says which stretch is right, and the answer swings between steep decline and growth.
Even a clean record has a limit, because a well is usually abandoned when its gas stops paying for itself: the price falls, the pipeline closes, or a repair costs more than the gas is worth. A decline curve fitted to the end of that life describes those decisions. Its forecast of what remains is, by construction, what was still worth selling under those conditions, which for an abandoned well is close to nothing. A leak works on different terms. It has no sales line to push against and no price to clear, so it can draw on gas a producer had already written off.
Even a clean fit therefore answers how production ended. It does not answer how much gas is left to leak.
What this means for anyone crediting these wells
A credit for plugging a leaking well rests on the leak carrying on for years had the well been left alone. For an orphaned well, a decline-curve answer to that question has to say which record it came from, whether that record belongs to this well and not its lease, and how far the answer moves when the fitting window moves. For most of Oklahoma's orphaned wells the exercise ends at the first of those, because there is no record. For most of the rest, it ends at identity or at the window.
That does not leave the question without an answer. It means the answer has to come from the well itself, what holds the gas in and what the failed fitting lets out, rather than from a sales history it mostly never had. We take that up in a companion note, how long a leak lasts.