The cockpit observation
Initial recognition and changing appearance
The 2001 investigation places AF3532 over the Coulommiers region at 13:14, flying at 39,000 feet with 24 passengers aboard. It records excellent visibility above an altocumulus cloud layer and a ground speed of 350 knots in its narrative account. The chief steward drew attention to a phenomenon on the aircraft’s left, which he initially described as a weather balloon. The captain thought of an aircraft turning at about 45 degrees of bank. The co-pilot moved behind the captain’s seat to observe it. Investigation, printed p. 2.
These first identifications show how the observers initially tried to relate the appearance to familiar airborne objects. The later judgement that it resembled nothing they recognised followed continued observation. An initial comparison is relevant to appearance, but it does not establish that the source was a balloon or aircraft.
Duboc described a dark red disc becoming lens-shaped, then apparently turning transparent and disappearing without a visible departure. The co-pilot described a dark brown form changing from thick to thin and eventually resembling an arrow. She allowed that it might have moved away so rapidly that the visual effect resembled disappearance. Her account also distinguishes sharp outlines at the beginning and end from blurred outlines during the intervening phase. Investigation, printed pp. 2–3; reproduced forms and drawings, scan pp. 8–12.
Agreement and differences
Both pilots reported an unfamiliar appearance on the aircraft’s left, changing form and an abrupt end to the observation. Their accounts differ on motion: the captain described a source stationary in space as the Airbus passed it, while the co-pilot described movement roughly parallel to the aircraft, with a turn before disappearance. Those descriptions imply different trajectories. A combined narrative in which all three witnesses observed an identical manoeuvre would conceal the distinction.
Duration also varies. The investigation’s narrative says one minute, its discussion refers to one or two minutes, the captain’s reproduced form gives one or two minutes, and the co-pilot’s form gives two to five minutes. These are recorded estimates rather than a continuously timed visual observation. They matter because reconstructed distance depends on how far the Airbus travelled while the bearing changed. Investigation, printed pp. 3–4 and reproduced forms, scan pp. 8 and 10.
The third observer’s independent statement was not recovered by the investigators. The 2001 report explains that the steward could not be identified after Air France’s limited retention of crew-composition records. His initial reaction therefore survives through other participants and reporting documents. Three reported observers should not be treated as three equally complete, independently preserved accounts. Investigation, printed p. 2.
Both reproduced pilot forms mark no radar contact and no electromagnetic effects on aircraft systems. The claimed technical corroboration concerns ground air-defence reporting. The forms supply no onboard instrument detection of the visual source or recorded aircraft-system disturbance. Pilot forms, scan pp. 8 and 10.
Reporting and the documentary sequence
A report during the flight
The crew contacted the Reims air navigation control centre, CRNA-Est. The report was relayed to the Centre d’Opérations de la Défense Aérienne, CODA, at Taverny. A controller’s SEPRA form dated 28 January 1994 records a bell-shaped appearance becoming a flattened lens, with an estimated position to the left and below the aircraft. This provides a written reporting trace from the event date, independently of the later formal crew statements. Controller form, investigation scan pp. 22–23.
A surviving CODA chronology distinguishes the times at which information was received: 13:16Z, 13:20Z and 13:26Z. Its first entry places the sighting at 13:14Z and gives a duration of one minute. The later entries add a dark bell shape, an estimated distance of 10–20 nautical miles, a position 2,000 feet below, and a brown lens form. The fax header is dated 1 November 1995. It is a later transmission of an event-day information chronology, rather than a newly collected sighting. CODA chronology, p. 1.
The reception times do not extend the observation to 13:26. Information about a short encounter can arrive in successive messages after it has ended. Keeping observation time, report time and transmission date separate prevents an administrative sequence from becoming an invented twelve-minute visual track.
The controller form’s field for the start of observation contains 13:27 UTC, with 13:43 entered as the report time. This differs from CODA’s 13:14Z sighting entry and its earlier reception sequence. The inspected documents do not explain the discrepancy. Their agreement about the event date and reporting occurrence is firmer than an exact alignment of every written time. Controller form, scan pp. 22–23.
Formal testimony and investigation
The SEPRA report identifies 27 February 1997 as the date of the first written crew testimony, after the case entered press coverage. It records subsequent interviews and enquiries in 1999–2000; the synthesis is dated 10 April 2001. Investigators expressly recognised the possibility of retrospective reconstruction, including estimates becoming more internally consistent through repeated description. Investigation, printed pp. 2 and 4.
A report made during the flight establishes that the incident was not first invented in response to publicity three years later. The delay still affects the precision of later drawings, distance estimates and recalled movement. The event-day controller record and later interview material therefore serve complementary functions: one anchors the reporting event, while the other supplies detail with a greater interval for memory and discussion to affect it.
The captain’s letter of 27 February 1997 estimates a size of about 1,000 m and a thickness of 100 m; the later reproduced pilot form gives 200–300 m. His letter also says that the Paris Match article informed him of the radar signature. These details make the development of the account traceable: dimensions changed, and the participant’s knowledge of the proposed radar corroboration entered his written narrative through later publicity. Neither point invalidates the initial report, but both affect how independently the later interpretation can corroborate the technical record. Captain’s letter, Gendarmerie file PV 00351/97, scan p. 4.
What the radar information establishes
An unassigned track
The investigation describes a track from the Cinq-Mars-la-Pile control centre, moving north-east to south-west and crossing the Airbus’s projected ground path near Meaux. It gives approximately 180 km/h as the track’s ground speed and says that no corresponding flight plan was identified. The height-finding radar did not detect the target, so its altitude was unavailable. A low-altitude light aircraft remained one of the report’s possible explanations for the weak, brief return. Investigation, printed p. 5.
The source describes 50 seconds of recorded track within an interval from 13:13:50 to 13:15. Those endpoints span 70 seconds. The stated detection duration and the broader interval may concern intermittent returns, but the published account does not resolve that relationship precisely. The figures should retain their separate descriptions rather than being presented as one uninterrupted 50-second recording with those exact endpoints.
Why temporal coincidence is insufficient
The investigation’s introductory account says that the radar track and visual phenomenon disappeared at the same moment. Its analysis ends by cautioning about their correlation. The reported track crossed from the aircraft’s right to its left at comparatively low speed, while the crew placed the visual phenomenon much farther to the west on the left. The annexed comparison diagram illustrates the difference between the near radar path and the more distant estimated visual location. Investigation, printed p. 5 and annex diagrams, scan pp. 13–14.
Without target altitude, the diagram’s crossing of ground projections does not establish a close approach at the Airbus’s height. Nor can radar timing determine the visual source’s distance unless their common identity has first been demonstrated. The combined hypothesis needs a trajectory compatible with the sightlines, time uncertainty and detection characteristics. Simultaneous loss of visibility and a weak radar track can be relevant evidence, but the proposed correspondence remains a question to test.
Distance, size and quantitative reconstruction
Conditional geometry
The captain’s later estimate placed the phenomenon at 25 nautical miles. SEPRA tested a stationary source seen from ten o’clock to eight o’clock during two minutes while the Airbus travelled at 330 knots. The resulting construction gave a distance of roughly 11 nautical miles, or approximately 14 nautical miles under a wind-carried-source assumption. Maintaining a 25-nautical-mile distance required different viewing angles or assumed source motion. Investigation, printed p. 4 and calculation annex, scan p. 15.
This is a consistency test of recollected inputs. Its 330-knot assumption differs from the narrative’s 350-knot figure, and the source supplies several duration estimates. The model is useful because it exposes dependence on those assumptions. Its alternative distances cannot acquire the status of independently measured range simply because they were calculated.
An angular-unit inconsistency
The size discussion describes an apparent diameter of half the Moon’s diameter, then prints “15 sec. d’arc”. It gives a scale of 0.40 m at 100 m and resulting sizes of 83 m at 11 nautical miles and 190 m at 25 nautical miles. These scales are incompatible with 15 arcseconds: 0.40 m at 100 m represents an angle of approximately 14 arcminutes. The calculations are of an arcminute order, despite the printed arcsecond label. Investigation, printed p. 5.
That internal inconsistency calls for checking the intended unit before reusing the calculation. More fundamentally, even a corrected angular unit would leave physical size dependent on an unmeasured distance. The record supports a reported large apparent form, with conditional size estimates. It does not provide an instrumented measurement of a vehicle hundreds of metres across.
The weather-balloon enquiry
Météo-France’s note dated 9 January 2012 examined a Trappes radiosonde launched at 11:18 UTC, explicitly given as 12:18 local legal time. It describes a balloon, parachute and VAISALA RS80-15L radiosonde located through Loran C, with no radar tracking reflector of the type previously used at Trappes. Its plot contains a data-reception gap: the note warns that a straight segment joins known endpoints rather than reproducing the actual intervening path. Météo-France note, p. 1.
The note places the assembly near Joigny at 17,404 m after 56 minutes, labelling that position 13:14. It gives bursting after 90 minutes at 31,411 m, labelled 13:48. Those elapsed times correspond to the local-time launch of 12:18, whereas the CODA observation is explicitly 13:14Z. Comparing the balloon with the aircraft therefore requires a common UTC clock and the balloon’s later state. GEIPAN’s case narrative describes the balloon as already burst and descending farther south-east at the relevant observation time. Météo-France note, p. 2, GEIPAN narrative.
Météo-France also argued that a balloon could not reach the several-hundred-metre size attributed to the phenomenon. That argument uses a reported physical size whose distance basis is uncertain. A robust test of a specific balloon candidate needs its angular appearance and sightline from the aircraft at the correct time. The enquiry weighs against an uncomplicated identification with the named Trappes launch; it cannot exclude every wind-carried object solely by comparing balloon size with the crew’s derived dimensions.
Documentary chronology
- 28 January 1994: The crew report an observation at approximately 13:14 UTC; controller and CODA records preserve the reporting sequence.
- 1 November 1995: The surviving CODA chronology carries a fax transmission date, distinct from the event it describes.
- 27 February 1997: SEPRA dates the first formal written crew testimony to this day, following renewed press attention.
- 1999–2000: Investigators conduct further enquiries and interviews; the captain’s reproduced drawing is dated 14 June 2000.
- 10 April 2001: The SEPRA investigation synthesis is dated.
- 9 January 2012: Météo-France dates its Trappes radiosonde note.
Evidence assessment
The event-date controller form and CODA information chronology anchor the aviation report. The later crew accounts give substantial descriptive material but retain differences in duration, motion and appearance. The radar information adds a separate technical record whose identity and altitude are unresolved. These distinctions permit a serious assessment without treating professional experience as a substitute for measured range or paired sensor data.
GEIPAN’s detailed narrative describes the case as D1, while its page header displays the broader D label. The institution’s category records an unresolved assessment. The visible difference between header and narrative does not itself establish a new physical finding or a documented substantive reclassification. The assessment of individual claims rests on the records rather than the category alone.
A balloon or other wind-carried object
The initial balloon comparison and changing outline make a flexible airborne object a relevant candidate. The named Trappes radiosonde has specific geographical and timing constraints. Evaluating another such source would require a documented launch or independently reconstructed position; a resemblance alone cannot establish identity.
Conventional traffic and a separate visual source
SEPRA retained a low-altitude light aircraft as a possible source of the weak radar return. The missing radar altitude and differing trajectories permit the radar and visual observations to have separate causes. A successful explanation would identify each component rather than assuming that explaining the return resolves the cockpit appearance.
A common unidentified source
A common source remains possible if a reconstruction can reconcile the visual bearings and radar trajectory. The abrupt visual disappearance and reported shape change need assessment through viewing geometry and visibility before they can support extraordinary performance. The available record leaves physical origin unresolved.
Assessment snapshot
| Proposition | Assessment | Basis and limit |
|---|---|---|
| An unfamiliar phenomenon was reported during the flight | Well supported as a reporting event | Event-date controller form and CODA information chronology |
| Three observers supplied equivalent independent accounts | Uneven documentary support | Two later pilot accounts survive; the steward’s independent statement was not recovered |
| Radar confirmed the visual source’s identity and altitude | Unresolved association | Different trajectory information and no radar target altitude |
| The source measured hundreds of metres across | Conditional estimate | Unmeasured range and an angular-unit inconsistency in the size calculation |
| The Trappes balloon enquiry settles all balloon explanations | Excessive scope | A specific candidate was studied; common-clock geometry and uncertain physical size remain material |
Claim ledger
| Claim | Source function | Evidence that would change the assessment |
|---|---|---|
| A brown or dark red form changed appearance and disappeared | Controller reporting and later participant accounts | Original audio and independently preserved sightlines or imagery |
| Visual disappearance coincided with loss of the same radar target | Investigation synthesis and proposed correspondence | Original time-stamped radar returns, detection thresholds and a compatible three-dimensional trajectory |
| A giant object was measured | Witness estimate and conditional calculation | Independent distance, a verified angular measure and corrected source units |
| A Trappes radiosonde was definitively excluded by size | Meteorological assessment using a reported dimension | A common-UTC balloon reconstruction compared with angular appearance from the aircraft |
Open questions
Can visual and radar positions be reconciled?
The common-source hypothesis needs more than temporal overlap. Original returns, aircraft positions and uncertainty ranges could test whether the proposed sightlines ever meet the radar trajectory at a credible altitude.
What did the earliest messages contain?
The surviving controller form and transmitted chronology provide an early reporting chain. Complete audio and operational records would distinguish the initial description from details added during later interviews.
Which balloon position matches the observation clock?
The 2012 note’s elapsed times identify local-time entries, while CODA uses UTC. The full trajectory at the aircraft’s observation time would permit a more reproducible test of the named candidate.
Further research
Witness record
The captain, co-pilot and chief steward were the reported cockpit observers. Duboc’s public interview is reproduced in the investigation’s press annex; personal identifiers in formal records are redacted. The steward’s missing independent statement remains a source limitation. No separate UAPRAD witness deep dive is currently mapped for this case.
Source library
Event reporting and participant records
- CODA chronology of information received on 28 January 1994: one-page record with a 1 November 1995 fax header; UTC reception entries remain distinct from observation time.
- Gendarmerie file PV 00351/97: administrative inventory and synthesis, scan pp. 1–3, and the captain’s letter dated 27 February 1997, p. 4. Redacted personal details remain redacted.
- Controller form and reproduced pilot forms in the SEPRA investigation: scan pp. 8–12 and 22–23. The controller form is dated to the event; participant drawings and forms belong to the later investigative record.
Institutional investigation and archive
- SEPRA/CNES investigation synthesis, 10 April 2001: printed pp. 1–6, followed by annexes. Printed and PDF scan numbering differ by one in the main report.
- GEIPAN case record: the institution’s current narrative, classification display and original attachment catalogue. Its compact radar-confirmation wording needs to be read alongside the detailed reservations.
Technical analysis and claim history
- Météo-France Trappes radiosonde note, 9 January 2012: pp. 1–2 contain the reconstruction and assessment; p. 3 contains the wind table.
- Press annex in the SEPRA investigation: scan pp. 24–26 preserve 1997 publicity and an attributed Duboc interview. These articles document public interpretation, rather than additional instrumented observations.
The linked PDFs are public archival copies hosted by UFO Transparency. Their original institutional titles and page contents provide the source basis here; the host’s generated summaries are not used as independent corroboration. Multiple reproductions of the same record represent one documentary chain.