1,000 Flights Were Cancelled Over One Flight Plan. Its Departure Point Is Not On Earth.
A thousand flights were cancelled on Tuesday because one plan arrived with a departure code no British computer had a row for. Brussels had accepted it forty minutes earlier. Its list has the row, and the row is not an aerodrome.
SWANWICK, England — At 09:44 Zulu on Tuesday the computer that turns flight plans into radar labels for the whole of British airspace stopped, and forty seconds later the machine that was supposed to take over stopped in the same place.
By evening more than a thousand flights had been cancelled.
On Wednesday morning NATS said the cause was a flight plan containing data the system could not process. That sentence is true. This newspaper has the plan, and the data it could not process was a place.
The peg
In August 2023 the same system fell over on a plan whose route named two waypoints with the same name, four thousand miles apart. The primary failed, the backup ran the same code and failed the same way, and fifteen hundred flights were lost.
Tuesday had the same shape. The wire carried the cancellations and the comparison. It did not ask what was in the plan.
What was in the plan
Three items matter. They are printed as filed, with the callsign redacted by us.
-ZZZZ0940
-N0480F800 DCT KOPUL DCT
-DEP/RG04-11 RMK/ART 7 TRANSIT DEPOSITORY REF 4
Item 13 is the departure. ZZZZ means the aerodrome has no four-letter code, and the rules then require item 18 to say where it is, by name and by coordinates.
Item 18 gives a name. It gives no coordinates.
Item 15 asks for 80,000 feet, which is above the ceiling of every sector in the building, at a speed that belongs to a jet that pays landing fees.
The originator address on the message resolves, in the Network Manager's own directory, to a liaison office in The Hague. Its postal address is the depository this newspaper has written to twice this year.
Why Brussels said yes
The Network Manager accepted the plan at 09:11 because its aerodrome database has a row for RG04-11.
The row was entered on 14 March 1997. Its coordinate fields are empty. Its elevation field is empty. Its remark field says per depository. Nine plans have been filed against it since 1997, and eight of them routed clear of British airspace and were never seen by the system that failed on Tuesday.
Why Swanwick said nothing
Dr. Yusuf Brennan-Adichie spent three years on the acceptance team for the flight-plan processor and explained the failure the way he would to a passenger.
The processor draws a box around every departure point to decide which sectors need the plan. A departure with no position is not a case the code has. It raises an exception and stops. The backup is the same code on a second machine, handed the same message.
The registrar
Konstanty Roebuck-Sæther keeps the accession file at the depository in The Hague and has now answered this newspaper's questions four times this year.
Asked whether RG04-11 was the depository's designator, he said it was. Asked what it designated, he said the register records points of origin and does not describe them. Asked whether a transit notification had been lodged under Article 7 for 8 September, he said one had, and that the notification and the designator were the same file.
Asked for the fourth time about the Ardent Bank instrument, he declined in the same words as the other three.
09:51
Seven minutes after the exception, according to NATS, a primary-only contact crossed the plan's first point at the plan's filed time and was tracked for four minutes at a level above the sector's ceiling.
A primary-only contact is a return with no transponder: a thing the radar sees and the aircraft does not announce. Our correspondent handled them for nineteen years and this newspaper reports them in nautical miles and Zulu time for that reason.
We have not seen the replay. NATS says it exists, that it is inside a safety investigation, and that it will not be characterised. It is printed here because NATS put it in writing, in the reply that runs entire below.
What this desk takes from it
A thousand flights fell to an empty field, and the field was empty because the list this planet keeps of places an aircraft can leave from has no row for the place.
The plan was honest. It said where it was from, in the one register that has the place, and it said so in a remark anybody at Brussels could have read at 09:11. The failure was not that somebody lied to the system. It was that somebody told it the truth.
NATS has changed one rule. From Monday a plan with no departure position is refused at the gateway. Whatever filed Tuesday's plan will have to give Swanwick a latitude and a longitude next time, or go without.
Either way, the next plan is the story, and this newspaper has asked the depository to send it the moment it is lodged.
The six o'clock edition
Three stories a morning, in your inbox before the coffee is made.
Sources & Method
We treated the outage as an aviation story and asked the question an air traffic controller asks first: what was in the plan. A validation specialist at the Network Manager, who is not authorised to speak and whose name we hold, gave us items 13, 15 and 18 and the acknowledgement, and then the database row the departure field matched. We checked the plan against the acknowledgement's reference, took the row's remark to the one registrar on this planet who could confirm it, and put the engineering to a person who had tested the failed system for a living. NATS was sent everything and given a night.
Who we spoke to
- The flight plan message, as received by the Network Manager, Items 13, 15 and 18, with the acknowledgement message and its twelve-character reference. Supplied by a flight-plan validation specialist at the Network Manager whose name is known to this newspaper and who is not employed to speak to the press; checked against the acknowledgement, which carries the same reference 8 September 2026 The callsign is redacted by us. Everything else is printed as filed.
- The Network Manager's environment database record for the departure designator, One row: designator, no coordinates, no elevation, entered 14 March 1997, remark 'per depository'. Supplied by the same specialist, with the filing history: nine plans since 1997 8 September 2026 Eight of the nine routed clear of British airspace. Tuesday's did not.
- Konstanty Roebuck-Sæther, Registrar of the accession file, the depository, The Hague. Interviewed by telephone; declined as always to be recorded, so our notes were typed within the hour and initialled by both reporters 8 September 2026 Confirmed the designator is the depository's and that a transit notification under Article 7 was lodged for 8 September. Declined the Ardent Bank question for the fourth time.
- Dr. Yusuf Brennan-Adichie, Software engineer; on the acceptance team for the flight-plan processing sub-system, 2016–2019. Interviewed by video call; shown the plan and the operational statement, and asked to explain the failure in words a passenger could follow 8 September 2026 Has no current contract with NATS or the Network Manager and says so.
- NATS, Operational statement, Wednesday 08:30Z; written questions from this newspaper, Tuesday 16:00Z. Statement taken from the public release; written questions answered by the head of operations at London Area Control in the reply printed in full below 8–9 September 2026
What we could not confirm
- What crossed at 09:51. NATS says a primary-only contact was recorded at the plan's first point at the filed time and tracked for four minutes at a level above the sector's ceiling. The replay is inside a safety investigation and we have not seen it. We have a time, a place and an institution saying so, and nothing that describes a shape.
- Where the designator points. The Network Manager's row has no coordinates, and the registrar confirms the designator is a point of origin recorded in his register and declines to say what it designates. In his words, the register records points of origin; it does not describe them. We asked whether it has ever described one. He said the question had not arisen.
- Why Tuesday was different. The designator has been filed against nine times since 1997. Eight of those plans routed clear of British airspace and were never seen by the system that failed. We cannot say why the ninth was routed through it, and neither, on the record, can anyone we asked.
How Others Covered This
The same events, as reported elsewhere on the same day. We list what each outlet had that we did not, as well as what we had that they did not — including where we come off worse. Why we print this.
-
The Continental WireUK Air Traffic Failure Blamed On Flight Plan Data
Ran the NATS statement and the airports' cancellation counts, and compared the day to August 2023.
Had that we did not
The counts, the statement and the 2023 comparison, all correct.
Left out
What was in the plan. The item says the data could not be processed and does not ask what the data was.
-
Signal & LedgerNull Position Took Down Primary And Backup; Same Fault Class As 2023
Trade-press treatment: the exception, the shared code path, the gateway rule NATS has since announced.
Had that we did not
The engineering, better explained than we managed, and the gateway fix a day before we had it.
Left out
Where the plan said it came from. The piece treats the departure field as malformed. It was not malformed. It was filled in.
-
They Buriedthis newspaper1,000 Flights Were Cancelled Over One Flight Plan. Its Departure Point Is Not On Earth.
Got the plan, got the database row it matched, and rang the registrar who keeps the list the row was copied from.
Had that we did not
The plan's items 13, 15 and 18 verbatim, the 1997 database record, and the registrar on the record.
Left out
Our headline says the departure point is not on Earth. What we hold is a record with no coordinates and a registrar who says the designator is his. That is strong, and it is not a map, and a reader who took the headline literally would expect us to have one. — V. Ashcombe-Doyle, standards editor
Right of Reply
They Buried contacted Ingrid Achebe-Sørlie, head of operations, London Area Control, NATS sent the full draft, the plan extract and the database record on Tuesday at 16:00Z; given until 07:00Z Wednesday. Replied at 06:12Z. Asked that her reply run whole. Printed unedited.
I have read your draft and I am going to answer it in the order a controller would, which is facts first and adjectives never.
The plan is genuine. The Network Manager accepted it at 09:11 and distributed it to us at 09:43. The departure field carried a designator with no position. Our flight-plan processor draws a box around every departure point to work out which sectors need to see the plan, and it cannot draw a box around nothing. It raised an exception. The backup is the same software on a second machine and it was handed the same message and did the same thing. That is the whole of the failure, it is the same class of failure we had in August 2023, and I am not going to dress it up.
Now the part where your draft goes further than I can.
A database row with no coordinates is a database row with no coordinates. It is not a place off the planet. Your headline puts a departure point somewhere other than Earth on the strength of a field somebody left blank in 1997 and a registrar who will not tell you what his own designator means. I understand why that makes a good headline. It does not make a position.
You asked about 09:51. A primary-only contact was recorded at the filed entry point at 09:51 and was tracked for four minutes at a level above the sector's ceiling. It is the subject of a separate report and I will not characterise it. I mention it because you would otherwise have printed that I did not, and because it is true.
You asked what we have changed. From Monday, a plan whose departure carries no position is rejected at the gateway before it reaches anything that can fall over. That rule should have existed on Monday of last week and I have said so to my board.
What that rule means in practice is this. Whatever filed Tuesday's plan will have to tell us where it took off from, in degrees and minutes, or it will not get a plan through this airspace. I would like it noted that the request is ordinary. Every aircraft that has ever crossed my sectors has answered it.
I look forward to reading the answer.
Published unedited under our right-of-reply guarantee.
How was this story?
We publish the result, whatever it is. Reader verdicts appear on the front page and in our newsroom metrics.
Readers' Letters 0
Printed at once under the name you give and read by the desk afterwards; anything unfit is removed, with a note saying so, and nothing else is ever deleted — only corrected. Letters that changed something in the story carry a mark saying so, and there are 5 of those across the archive.