On 6 September 1998, an unfamiliar object seen above a tree at Voreppe, in Isère, became the subject of witness statements, a camcorder recording and a CNES investigation. The eventual identification was an escaped helium balloon shaped like a ladybird. The surviving archive is particularly useful because it preserves several stages of the inquiry: initial descriptions, an announcement that identification was still pending, the identification itself, and a later technical examination of the recording.
Those stages answer different questions. A witness can describe an appearance accurately without recognising its cause. A video can confirm that something was visible without providing enough detail to identify it. An explanation can become convincing through information obtained outside the camera frame. Voreppe brings those distinctions together unusually clearly, without requiring either the initial testimony or the subsequent explanation to be discarded.
The first descriptions
Statements taken on the evening of 6 September and the following morning describe a silent object above a neighbouring tree, rocking before moving away towards the Vercors. One observer compared its appearance with a large ladybird. Another described six leg-like projections, a dark body and a red band, but expressly said she had not seen portholes. The camcorder operator estimated a much larger object and recorded its departure at approximately 20:02. These are estimates and interpretations in police testimony, not independently measured dimensions.
The differences matter. Window-like markings in one account do not become confirmed windows when another observer denies seeing them. Likewise, projections can be remembered as mechanical legs while belonging to a familiar decorative shape. The statements preserve uncertainty within the original observation, rather than a single agreed description of an engineered vehicle.
The released drawings add a visual counterpart. Selected sheets show an oval underside, a raised upper section and leg-like appendages; a separate reconstruction places a drawn outline above a tree photograph. These are witness representations and reconstructions, not photographs of the object at that apparent size. Their value lies in recording the observers’ visual interpretation. Reproducing a reconstruction without its label would blur the very distinction the archive allows readers to examine.
Two announcements, one day apart
The first CNES press statement, dated 11 September, said SEPRA had attended on 8 September after contact from the gendarmerie. Meteorological and aviation information was being collected and the recording examined. The agency explicitly withheld a conclusion about the object’s nature or origin. On 12 September, a second statement announced the identification: an animal-shaped helium balloon, approximately a metre across. It attributed the unfamiliar appearance to the combination of sunset, light wind and an urban background.
The first announcement therefore cannot fairly be presented as the agency’s final verdict. Nor should the second be backdated into the initial response. The short interval demonstrates an investigation changing its answer as information arrived. Both statements belong in the history, with their dates attached.
GEIPAN’s present case account supplies the identification’s practical link. A parent who learned of the affair through the press reported that a ladybird-shaped balloon, with conspicuous legs, had slipped from a child’s wrist. Its escape matched the observation’s sector and time. The record now carries category A. This account concerns an identifiable source for the reported appearance, rather than a balloon suggested solely because it could theoretically have been in the sky. The published case summary reports the matching testimony; it is not a substitute for an independently recovered balloon.
Weather and an aviation record
The original Météo-France sheet gives conditions for 20:00 local time on 6 September at Grenoble–Saint-Geoirs airport. It records a north-easterly wind of 2 metres per second, visibility of 50 kilometres and three eighths cirrus cloud cover. This is a dated observation at a named station. It should not be turned into an exact measurement of airflow beside a particular tree or between nearby buildings.
The aviation attachment also requires care. Its covering sheet forwarded an observation form to CNES; the completed form describes a primary radar track from Lyon–Satolas, with timed positions and a recorded disappearance. The form says its information came from reading the radar recording. In the reviewed material it does not establish that this return was the small object seen over the garden. Its presence in the same archive cannot supply that missing association.
A radar observation and a visual observation need a demonstrated connection before they can corroborate each other. Similar timing alone is insufficient: location, movement and identification also need to fit. This case is therefore not tagged as a radar-confirmed balloon sighting. The form remains part of the investigative record, without being recruited as evidence for a link it does not itself demonstrate.
What the video analysis could establish
Fleximage’s December 1998 report was written after the balloon identification and treated the recording as a reference case. It analysed digitised VHS-C footage, selected 137 images and addressed interlacing and camera movement. The object occupied only a small number of pixels, limiting its recoverable detail. Its distance and speed estimates depended on assumed balloon size and camera settings, rather than independently measuring the object’s physical dimensions. The report considered the movement compatible with the balloon explanation, but expressly said that the identification could not be deduced from the video alone. It also treated an apparent final downward acceleration as potentially caused by the operator raising the camera while stopping the recording.
This qualification makes the technical report more useful, not less. Once an object has been identified through other evidence, an analyst can ask whether its recorded appearance is consistent with that identification. That exercise differs from giving an analyst an unknown image and obtaining a unique answer. Confusing the two would exaggerate what the recording resolved.
The distinction also protects against treating every apparent movement as the object’s motion. A hand-held camera records changes in its own direction as well as changes in the scene. Features used as references can help, but they do not remove every uncertainty. Numerical outputs derived from an assumed size remain conditional even when the calculation is careful.
Assessment and locality
The supported outcome is explained: CNES announced a balloon identification, and GEIPAN records the matching account of an escaped ladybird balloon. This conclusion rests on the identification evidence together with the compatible appearance and recording. It does not depend on claiming that each witness estimate was correct, that the video alone proved the balloon’s identity, or that the separate radar return belonged to it.
The surviving documents also show why an initially puzzling report need not imply deception. Reports of silence, apparent legs and an unusual silhouette can preserve real perceptual details while assigning them an unfamiliar meaning. Here the useful historical question is how those details were connected to an ordinary object, and which parts of that connection came from testimony, investigation or later analysis.
The map uses an approximate Voreppe town reference at 45.298, 5.637, rounded from sourced commune coordinates. It does not identify a witness’s garden or the balloon’s release address. The lead photograph shows Grande Rue in June 2023 and provides modern locality context only. Neither the photograph nor the map purports to reconstruct the evening’s exact viewpoint.
Sources and further reading
GEIPAN: current case account and matching escaped-balloon testimony
Gendarmerie: released statements and drawings, PV1684/98 (selected sheets reviewed)
CNES: successive press statements, 11 and 12 September 1998 (both read)
Météo-France: conditions at Grenoble–Saint-Geoirs, 20:00 local time, 6 September 1998
Fleximage: December 1998 video analysis (selected methods and conclusion reviewed)
Released witness drawings and reconstructions (selected sheets reviewed)
AirScott: exact modern photograph, creator and CC BY-SA4.0 rights