Roadmap · On the witness's side
What is known about witnesses.
What studies of observers' perception and memory establish, and what UFO@home should do with it. No source is an authority on its own: each lead says whether it is established by several independent studies, rests on a single study, is only argued, or is disputed because the sources disagree.
Perception#
Metres are not data; angles, counts and layout are established#
On re-entries seen by hundreds of people, stated distances and altitudes are off by factors of ten or more whatever the truth (Zond IV was placed from treetop level to 60,000 ft, for an object 100 miles up), while angular size, number of lights, layout and heading agree. Metres come back only when anchored: a known-size object, an occlusion, a triangulation between distant, independent observers.
Sources: Hartmann, Condon report, 1968, Rospars, 2023, Bullard, 2014, Hendry, The UFO Handbook, 1979, Besse, GEPAN technical note 2, 1980, Oberg, 1994.
Stated angular size disputed#
Asked to size the Moon at arm’s length, 238 people answered 7 cm on average for 0.5 cm, and a GEIPAN case drew a plane 30 times too big. But sizes shown by hand on the spot were within 2 to 17 %, only a size recalled through a familiar object being 30 % too large; and for aircraft, a “large” size was often the spread between the outermost lights. Asking the observer to show the full Moon the same way measures their own factor.
Sources: Peiniger, 1998 (in 2023), Passot, 2023, Keul, 2023, Hartmann, Condon report, 1968, Hendry, The UFO Handbook, 1979.
Elevation is overestimated, and low lights are placed at the horizon established#
Above 30 to 45°, an object is often said to be “overhead”; in 196 identified astronomical cases, estimates of 25-40° were at 10° or lower four times in ten. A light below about 5° is placed at the distance of the horizon or obstacle it overhangs, even the full Moon: a setting Venus “landed at the school”. The terrain model gives that distance, which makes such a statement testable.
Sources: Wertheimer, Condon report, 1968, van Utrecht, 2023, Jimenez, GEPAN technical note 10, 1981, Maugé on Jimenez, 2023, Hartmann, Condon report, 1968, Hendry, The UFO Handbook, 1979.
Direction across the line of sight, and its sense established#
A distant track is seen crossing the line of sight, so its heading is reliable within about 45°, and two observers contradict each other only when they give opposite directions. Where each observer stood predicts the sense: south of a fireball it passes left to right, north of it right to left.
Sources: Rospars, 2023, Hartmann, Condon report, 1968, AMS fireball #2908 (rr0.org), Hendry, The UFO Handbook, 1979.
Motion that is not motion established#
A lone light on a dark sky soon seems to move; an aircraft coming straight on seems to hover; a light that goes out seems to leave; a rocket brightening as it recedes seems to approach; a star “follows the car”. Each has a geometric test once the candidate is placed in the scene.
Sources: Wertheimer, Condon report, 1968, Rhine, Condon report, 1968, Hendry, The UFO Handbook, 1979, Oberg, 1994, Borraz, 2023.
Lights are joined into a body established#
A string of points is seen as lights or windows on a dark hull: Zond IV witnesses described the cigar with lit windows of the 1948 Chiles-Whitted sighting, and the 1913 fireball procession was already “two tiers of lights”. A “windows” report does not rule out a fireball; the question is whether a body was seen between the lights.
Sources: Hartmann, Condon report, 1968, Bullard, 2014, Hendry, The UFO Handbook, 1979.
The naming word disputed#
Jimenez found that the stronger the word (light, object, machine, UFO), the more precise and the more wrong the metric details, and that an experimenter saying “UFO” raised such labels from 18 % to 68 %. A study of 300 re-entry reports did not reproduce the link with distance. What is agreed: the longest, most detailed reports are not the most accurate.
Sources: Jimenez, GEPAN technical note 10, 1981, Maugé on Jimenez, 2023, Rospars, 2023, Hartmann, Condon report, 1968.
Colour disputed#
The main colour of a re-entry was consistent for 77 % of observers, yet witnesses of a 1962 meteor shower gave almost every colour; a dark-adapted eye infers the colour of a dim light, and aircraft green is often reported blue.
Sources: Rospars, 2023, Smith, The UFO Enigma, 1983, Wertheimer, Condon report, 1968, Hendry, The UFO Handbook, 1979.
Time#
Time, date and duration established#
The date is right for 98 %, the time within minutes for nine observers in ten, the duration within a factor of three. A clock read at the time beats every estimate: at Yukon in 1996 estimates spread over three hours, while the two who looked at a clock matched the re-entry. Duration is also the variable that selection distorts least.
Sources: Rospars, 2023, Peiniger, 1998 (in 2023), Bullard, 2014, Besse, GEPAN technical note 2, 1980, Hendry, The UFO Handbook, 1979.
A wrong date argued#
When nothing fits, the Moon or a planet often does a day, a month or a year off: a Polish case dated 1980 was 1979, and its “ball” vanished at the computed moonset behind a crest. Shifting the date, the hour, summer time and the direction is a check of its own.
Sources: van Utrecht, 2023, Borraz, 2023, Hendry, The UFO Handbook, 1979.
Memory and influences#
Retelling and delay established#
Each questioning lowers the ratio of correct to wrong details (from 3.5 to 1.39 between the first and the second, in mock-crime experiments), and meteor witnesses were mostly wrong after a day or two. A first interpretation, once formed, resists correction. The earliest version weighs most.
Sources: Sharps, 2023, Smith, The UFO Enigma, 1983, Hartmann, Condon report, 1968.
Versions of an account established#
Sizes grow, directions reverse and dates move between the versions of one account and between its publications; the press gave zigzags that the witnesses deny. Every dated version should be kept and compared field by field.
Sources: Oberg, 1994, Bullard, 2014, Besse, GEPAN technical note 2, 1980, Carlson, 2023.
Recognition, and how questions are asked established#
People recognise pictures far better than they describe them (over 98 % right away). Open questions distort less than closed or leading ones, and an investigator’s expectations shape the answers. Hence the order: a free account, then a choice among rendered views, and the questions stored with the answers.
Sources: Shepard, 1968, Wertheimer, Condon report, 1968, Rospars, 2007, Bullard, 2014.
Media, expectation, hypnosis and feedback established#
Some Zond IV witnesses found the sight unusual only after television covered it, a wave can start from a star seen in binoculars, and being told a place is a “hotspot” turns dust into spacecraft. Hypnosis produces false memories, and confirming feedback raises confidence without raising accuracy. What the observer had read, watched or been told before each version is part of the record.
Sources: Hartmann, Condon report, 1968, Hendry, The UFO Handbook, 1979, Besse, GEPAN technical note 2, 1980, Sharps, 2023, French, 2023, Rhine, Condon report, 1968, Kelley-Romano & Bradfield Douglass, 2023.
Several observers#
Independence established#
A group watching one light converges on one story; three in four of Hendry’s identified cases had several witnesses and were misperceived all the same. N accounts are N confirmations only if the observers did not talk, were questioned apart and reported before the press. GEPAN already rated cases on number, independence and prior beliefs.
Sources: Rhine, Condon report, 1968, Hendry, The UFO Handbook, 1979, Hartmann, Condon report, 1968, Jimenez, GEPAN technical note 10, 1981.
How many should have seen it single study#
Reports scale with the population in sight times the duration: about one report per six million person-seconds for bright fireballs. The visibility footprint the tool computes gives that population, so a single observer of a large, low object over a town becomes anomalous for everyone else’s silence.
Sources: Hartmann, Condon report, 1968, Smith, The UFO Enigma, 1983.
The observer#
Profession disputed#
Pilots and astronomers are not shown to be more reliable: a renowned astronomer took jets for a slow orange light, and pilots are trained to read any light as an aircraft on a collision course. Against this, Haines defends pilots’ accounts, and Blue Book’s own figures gave more unknowns among its best-rated reports. What matters is whether the expertise fits the stimulus.
Sources: Hartmann, Condon report, 1968, Oberg, 1985, Oberg, 1994, Hendry, The UFO Handbook, 1979, Haines, 2023, Blue Book Special Report 14, 1955.
State and viewing conditions established#
Dark adaptation, fatigue, alcohol or medication, corrected vision, a recent bright light (afterimages), binoculars on a twinkling star, a window, a windscreen or a jet exhaust in between: each changes what reaches the eye, and each is a field the recording can hold.
Sources: Wertheimer, Condon report, 1968, Rhine, Condon report, 1968, Hartmann, Condon report, 1968, Hendry, The UFO Handbook, 1979.
Candidates#
What night lights usually are single study#
Of 1,024 night lights Hendry identified: stars and planets 35 %, advertising planes 22 %, aircraft 19 %, meteors and re-entries 11 %; the Moon 2 %, where a French study found it in one case in five. The rates depend on the place and the period, so they are priors to set per sample, not constants.
Sources: Hendry, The UFO Handbook, 1979, Maillot & Abrassart, 2023.
Signatures established#
Silence never rules out an aircraft (91 % of low advertising planes were not heard); re-entries are slower, flatter and longer than meteors, often heading east, and many lights in a line seen from hundreds of kilometres apart; launch plumes stay sunlit for up to an hour and are twisted by the wind into “manoeuvres”.
Sources: Hendry, The UFO Handbook, 1979, Oberg, 1985, Oberg, 1994, Hartmann, Condon report, 1968, Young, 2023.
An explanation must be possible too argued#
For the lights seen at Exeter to be a refuelling tanker, it had to be within a mile and fly at 10 mph: the explanation fails, not the account. Every candidate is tested for what the observer could resolve and for how long it stays in view, and none may deny an established fact such as a flight log.
Sources: Bullard, 2014, Rospars, 2007.
Scoring#
Missing data and selection established#
A third of one pilot catalogue has no date or time; brightness is missing from two Blue Book reports in three. Selecting cases on credibility biases profession and age, selecting on strangeness biases shape and speed, and the reporting channel decides who reports at all. Completeness is scored, and expert review is stored rather than used as a filter.
Sources: Printy, 2023, Blue Book Special Report 14, 1955, Besse, GEPAN technical note 2, 1980.
Explained cases as the control group established#
Identified reports come from the same people, for the same reasons, as unidentified ones. A feature is anomalous only if it differs from the same feature in explained cases: the evening peak, for instance, is the same for both.
Sources: Hendry, The UFO Handbook, 1979, Leduc, 2023, Besse, GEPAN technical note 2, 1980.
Lies and mistakes established#
Verbal cues of deception reach about 70 to 78 % on samples and never decide a single account, and a mistaken witness passes any lie test. Stated doubt and spontaneous corrections are signs in favour of an account, not against it.
Sources: Statement validity analysis (rr0.org), Martins, 2023, Hendry, The UFO Handbook, 1979.
To read, and not yet on rr0.org#
Several studies these leads depend on are known here only through summaries: Manuel Jimenez's thesis on witness perception (1994) and his GEPAN technical notes, Askevis-Leherpeux and Cavé's study of independence indicators between witnesses (GEPAN, 1981), the experiments of Hans-Werner Peiniger (1998) and Alexander Keul, Richard Haines's study of drawings (1979), the Saros operation on lunar misidentifications (2019), the Hynek UFO Report (1977) for the pilot figures, and Matthew Sharps's measurements of witness memory. Reading them in full is the next step, before any lead becomes a check. The complete list, with references, is in the repository.