Acerca de los fenómenos luminosos ransitorios observados en la noche.pdf

Resumen
Las antiguas imágenes astronómicas digitalizadas, tomadas antes de la era de los viajes espaciales tripulados, ofrecen una visión excepcional del cielo antes de la era de los satélites artificiales. En este artículo, presentamos las primeras búsquedas ópticas de objetos artificiales con alta reflexión especular cerca de la Tierra. Seguimos el método propuesto por Villarroel et al. y utilizamos una muestra de transitorios extraída de Solano et al. Empleamos imágenes del Primer Sondeo del Cielo de Palomar para buscar múltiples transitorios (dentro de una exposición de placa) que, además de ser puntuales, estén alineados a lo largo de una banda estrecha. Proporcionamos una lista de las alineaciones candidatas más prometedoras, incluyendo una con una significancia estadística de ∼3,9σ. Estos transitorios alineados siguen siendo difíciles de explicar con los fenómenos conocidos, incluso si no se puede descartar por completo la posibilidad de que fuentes puntuales produzcan imágenes fantasma ópticas poco frecuentes. Exploramos las posibilidades restantes, incluyendo reflexiones rápidas de objetos altamente reflectantes en órbita geosíncrona o emisiones de fuentes artificiales muy por encima de la atmósfera terrestre. También encontramos un déficit altamente significativo (∼22σ) de transitorios de POSS-I dentro de la sombra de la Tierra en comparación con la cobertura teórica de la sombra hemisférica a 42.164 km de altitud. El déficit persiste, aunque con menor significancia (∼7,6σ), al considerar una cobertura más realista basada en placas tectónicas. Este estudio debe considerarse una exploración inicial del potencial de los estudios fotográficos de archivo para revelar fenómenos transitorios, y esperamos que motive búsquedas más sistemáticas en conjuntos de datos históricos.

Conceptos del Tesauro Unificado de Astronomía: Búsqueda de inteligencia extraterrestre (2127); Detección de transitorios (1957);

Estudios (1671); Astronomía del sistema solar (1529)

1. Aligned,Multiple-transientEventsintheFirstPalomarSkySurvey
BeatrizVillarroel
1
,EnriqueSolano
2,3
,HichemGuergouri
4
,AlinaStreblyanska
5
aa ,StephenBruehl
6
,VitalyM.Andruk
7
,
LarsMattsson
1
,RudolfE.Bär
8
,JamalMimouni
9
,StefanGeier
5,10
,AlokC.Gupta
11
,VanessaOkororie
12
,KhaoulaLaggoune
13
,
MatthewE.Shultz
14
,andRobertA.Freitas,Jr.
15
1
Nordita,KTHRoyalInstituteofTechnologyandStockholmUniversity,HannesAlfvénsväg12,SE-10691Stockholm,Sweden
2
DepartamentodeAstrofísica,CentrodeAstrobiología(CSIC/INTA), POBox78,E-28691,VillanuevadelaCañada,Spain
3
SpanishVirtualObservatory,Spain
4
ScienceoftheMatterDivision.ResearchUnitinScientificMediation,CERIST,Constantine,Algeria
5
InstitutodeAstrofísicadeCanarias,AvdaVíaLácteaS/N,LaLaguna,E-38205,Tenerife,Spain
6
DepartmentofAnesthesiology,VanderbiltUniversityMedicalCenter,701MedicalArtsBuilding,1211Twenty-FirstAvenueSouth,Nashville,TN37212,USA
7
MainAstronomicalObservatoryObservatoryoftheNASofUkraine,27,AkademikaZabolotnohoSt.,Kyiv,03143,Ukraine
8
InstituteforParticlePhysicsandAstrophysics,ETHZurich,Wolfgang-Pauli-Strasse27,CH-8093Zurich,Switzerland
9
DepartmentofPhysics,UniversityofConstantine-1,LPMPS&CERIST,Constantine,Algeria
10
GranTelescopioCanarias(GRANTECAN), CuestadeSanJosés/n,38712BreñaBaja,LaPalma,Spain
11
AryabhattaResearchInstituteofObservationalSciences(ARIES), ManoraPeak,Nainital,263001,India
12
CenterforBasicSpaceScience,NationalSpaceResearchandDevelopmentAgency,Enugu,Nigeria
13
SiriusAstronomyAssociation,Algeria
14
DepartmentofPhysicsandAstronomy,UniversityofDelaware,USA
15
InstituteforMolecularManufacturing,PaloAlto,CA,USA
Received2025June8;revised2025September23;accepted2025September24;published2025October17
Abstract
Old,digitizedastronomicalimagestakenbeforethehumanspacefaringageofferarareglimpseoftheskybefore
theeraofartificialsatellites.Inthispaper,wepresentthefirstopticalsearchesforartificialobjectswithhigh
specularreflectionsneartheEarth.WefollowthemethodproposedinVillarroeletal.anduseatransientsample
drawnfromSolanoetal.WeuseimagesfromtheFirstPalomarSkySurveytosearchformultiple(withinaplate
exposure) transientsthat,inadditiontobeingpoint-like,arealignedalonganarrowband.Weprovideashortlist
ofthemostpromisingcandidatealignments,includingonewith∼3.9σ statisticalsignificance.Thesealigned
transientsremaindifficulttoexplainwithknownphenomena,evenifrareopticalghostingproducingpoint-like
sourcescannotbefullyexcludedatpresent.Weexploreremainingpossibilities,includingfastreflectionsfrom
highlyreflectiveobjectsingeosynchronousorbit,oremissionsfromartificialsourceshighaboveEarth’s
atmosphere.Wealsofindahighlysignificant(∼22σ) deficitofPOSS-ItransientswithinEarth'sshadowwhen
comparedwiththetheoreticalhemisphericshadowcoverageat42,164kmaltitude.Thedeficitisstillpresent
thoughatreducedsignificance(∼7.6σ) whenamorerealisticplate-basedcoverageisconsidered.Thisstudy
shouldbeviewedasaninitialexplorationintothepotentialofarchivalphotographicsurveystorevealtransient
phenomena,andwehopeitmotivatesmoresystematicsearchesacrosshistoricaldatasets.
UnifiedAstronomyThesaurusconcepts:Searchforextraterrestrialintelligence(2127); Transientdetection(1957);
Surveys(1671); Solarsystemastronomy(1529)
1.Introduction
Digitizedskysurveyshavebroadenedthetimewindowin
whichwecanstudychangesinthesky.Programssuchasthe
DigitalAccesstoaSkyCenturyatHarvard(DASCH;Grindlay
etal.2012), theDigitalSkySurvey
16
(DSS), theUkraineVirtual
Observatory(JDAUkrVO;Vavilovaetal.2012;Vavilovaetal.
2017), andCarteduCiel,provideimagesoftheskyspanningnot
justafewdecadesbut,insomecases,over150yr.
Whilephotographicplatesarenolongerusedforlarge
astronomicalsurveys—havingbeenreplacedbysignificantly
fasterandmoresensitiveCCDs—thearchivalimagesstill
serveimportantscientificpurposes.Forexample,theyallow
studiesoflong-termvariabilityofastronomicalsourcesover
timescalesofdecadesorevenacentury,assumingtheobjectis
brightenoughtobedetected.
Anotheruseofthesearchivesistosearchforvanishingstars
andothertransients.IntheVanishingandAppearingSources
duringaCenturyofObservations(VASCO;Villarroeletal.
2016, 2020) project,imagesoftheskytakenintheearly
PublicationsoftheAstronomicalSocietyofthePacific,137:104504(22pp), 2025October https://doi.org/10.1088/1538-3873/ae0afe
©2025.TheAuthor(s). PublishedbyIOPPublishingLtdonbehalfoftheAstronomicalSocietyofthePacific(ASP).
aaaaaaa
Originalcontentfromthisworkmaybeusedundertheterms
oftheCreativeCommonsAttribution4.0licence. Anyfurther
distributionofthisworkmustmaintainattributiontotheauthor(s) andthetitle
ofthework,journalcitationandDOI.
16
https://archive.stsci.edu/cgi-bin/dss_form/
1

2. 1950s,priortothefirstanthropogenicsatellite,arecompared
withmodernsurveystoidentifypossiblesourcesthatmay
havedisappeared.VASCOemploystwocomplementary
approaches:first,anautomatedprocedure(Solanoetal.
2022) thatsearchesdigitizedimagedatafromtheFirstand
SecondPalomarSkySurveys(POSS-IandPOSS-II) for
transients;andsecond,acitizenscienceproject(Villarroel
etal.2022b) wherevolunteersclassifypotentiallyinteresting
objects.TheseeffortsarefacilitatedbytheSpanishVirtual
Observatory
17
anditssoftwaretools.TheVASCOprogram
hasresultedinthecatalogingofmanythousandsofunknown
transients,visibleonlywithinasingleplateexposure(Solano
etal.2022; Villarroeletal.2022b).
AnintriguingfindingfromtheVASCOprojectwas
presentedinVillarroeletal.(2021): ninefaint,star-like
objectsthatappearedandvanishedsimultaneouslyona1950s
POSS-Iplate.Thecentraltransientwasfirstidentifiedduring
thevisualvettingof24,000candidateswithsmallcutout
imagesoffixsize,yielding∼100vanishingstar-candidates,
seeTable2ofVillarroeletal.(2020). Inasubsequentfollow-
upofthese100transients,theimagecontainingninetransients
wasdiscovered.Theninetransientswerenotvisibleon
anotherplatetakenhalfanhourearlier,noronathirdplatesix
dayslater.Allknownastrophysicalexplanationswere
consideredbutdeemedimplausible.Thesurfacedensityof
suchtransientswastoohightobeattributedtoanyknown
naturalphenomenon.Whetherthiswasduetounknown
contaminationontheplatewithcoincidentallystar-likedefects
oragenuineastronomicalobservationremainsunresolved.If
real,oneexplanationcouldbethattheywerecausedbysolar
reflectionsoffflat,highlyreflectiveobjectsingeosynchronous
orbit(GSO) aroundtheEarth.Nevertheless,theninetransients
ontheirownareambiguous,particularlysincetheyarelocated
neartheplateedge,wheredefectsareknowntoaccumulate
(Hambly&Blair2024) andwhereotherroundobjectscanalso
befound.Yetthefindingpromptsaprovocativequestion:
couldsomeoftheobjectslongdismissedasplatedefects
actuallyrepresentreflectionsoremissionsfromartificial
sources?And,moregenerally,canpre-Sputnikplatesserve
asasearchdomainfornon-humanartificialobjects?
Findingsuchobjectsinpre-Sputnikdata,wouldrepresenta
significantdiscoverywithfar-reachingimplicationsforboth
astronomyandhumanity,includingthepossibilityofnon-
terrestrialartifacts(NTAs). Italsobearsdirectlyonthe
scientificinvestigationofUnidentifiedAnomalousPhenomena
(UAPs), formerlyknownasUnidentifiedFlyingObjects
(“UFOs”)—a subjectthat,afterdecadesofstigma,isnow
gainingseriousacademicattention,ashighlightedintherecent
reviewbyKnuthetal.(2025) inProgressinAerospace.
Clarifyingtheoriginofthesetransienteventsisthereforenot
onlyofastrophysicalinterestbutalsoofpotentialimportance
foroneofthemostenigmaticandconsequentialquestions
facingsciencetoday.
Toaddtotheintrigue,Solanoetal.(2023) recentlyreported
abrighttripletransienteventoccurringon1952July19,found
amongasetof∼5000short-livedPOSS-Itransients(Solano
etal.2022). Thishighlycurateddataset,inwhichdiagnostics
basedonphotometryandmorphometricparametershavebeen
carefullyappliedtothesampletoreducefalsepositives(e.g.,
platedefects), suggeststhatthephenomenonofmultiple
transientscanbefoundevenwhenstringentdiagnosticcriteria
areapplied.Asintheearliercasewiththeninetransients,the
objectsappearedandvanishedwithinasingle50minutes
exposure.Theirbrightness(r∼15–16mag)makescontam-
inationlesslikely.Notably,thisparticulareventcoincidesin
timewithoneofthemostextensivelydocumentedaerial
anomaliesinhistoricalrecords:theWashingtonD.C.“UFO
flap”of1952July,whichunfoldedovertwoconsecutive
weekends(July18–19and26–27). Whilethismaybea
coincidence,thetemporalproximityinvitesfurtherscrutiny—
especiallygiventherarityofbothphenomena.Inaseparate
studyBruehl&Villarroel(2025), weinvestigatepossible
statisticalassociationsbetweenhistoricalUAPreportsand
VASCOtransients,andfindpreliminaryevidenceofa
temporalcorrelationatthe∼3σlevel.Whilesuchafinding
doesnotimplycausation,itraisesthepossibilitythatcertain
anomalousaerialobservationsrecordedinthepre-satelliteera
mayhavehadphysicalcounterpartsobservableindeep-sky
imaging.
Giventheunusualnatureandpotentialimplicationsofthese
events,itisessentialtotestthehypothesisthatsometransients
arisefromreflectiveartificialobjectsinEarthorbit,andthat
certainpoint-likefeatureslongdismissedasplatedefectsmay
infactbesolarreflectionsfromartificialsurfaces.Searchesfor
extraterrestrialprobeswereproposedasearlyasthe1960s
(Bracewell1960), buttodateonlyafewsearchesforNTAs
havebeenattemptedorproposed(Freitas&Valdes1980;
Valdes&Freitas1983; Freitas&Valdes1985; Haqq-Misra&
Kopparapu2012).
Inapreviouswhitepaper,Villarroeletal.(2022a) proposed
amethodologytosearchforsolarreflectionsfromartificial
objectsinGSOusingphotographicplatesfrombeforethe
satelliteera(pre-1957). Onekeysignatureisthepresenceof
severalpoint-liketransientsthatarealignedalongalinewithin
asingleexposure.Astatisticalframeworkwasalsodeveloped
toassessthesignificanceofsuchalignments.
Inthispaper,wecarryoutthattest.Weapplythepublished
methodologyandstatisticalframeworktoapublishedsample
ofPOSS-ItransientsfromSolanoetal.(2022).
Weidentifyseveralpromisingcandidatesandexaminethem
indetailinSection5.Assumingtheeventsarereal,weusethe
alignedtransientstoinferthepossiblegeometryandsurface
densityofreflectiveobjectsnearGSO.Wealsoperforma
statisticaltesttoevaluatewhethersunlightisrequiredto
17
http://svo.cab.inta-csic.es
2
PublicationsoftheAstronomicalSocietyofthePacific,137:104504(22pp), 2025October Villarroeletal.

3. producethesetransients,basedontheirdetectionratewithin
Earth’sshadow.Finally,wediscussprospectsfordetecting
similarobjectsinmoderndigitalskysurveys.
2.PlateDefectorTechnosignature?
Oneofthecorechallengesinourworkisthecontamina-
tionofphotographicplatesbyartifactsthatmaymimic
astronomicalsources.Apparenttransienteventsinthese
platesoftenpresentacaseofdegeneracy—wheregenuine
astrophysicalsignalsandmundanedefectscanappear
strikinglysimilar.Certainplatedefectsareknownto
resemblestellarprofiles(Greineretal.1990), anddistin-
guishingthemfromauthenticobservationsremainsanon-
trivialtask,evenwhenfull-width-half-maximum(FWHM)
comparisonsareapplied.Moreover,defectscanclusternear
plateedges,andvignettingorunevendevelopmentmay
furtherconfoundinterpretation.Nevertheless,visualinspec-
tionandphotometricprofileanalysisremainindispensable
toolsinthisearlyphaseofexploration.
Rigorousdiagnosticswithquantitativemeasurementsare
centraltoanysearchforgenuinetransientsinphotographic
plates,asoverlypermissivecriteriainevitablyadmitlarge
amountsofnoise.Forthatreason,weshallusecarefully
selectedtransientsamplesinSolanoetal.(2022), which
average167transientsperplateandhavebeenmatchedto
severalmodernsurveystoremovevariablestars,asteroids,and
comets.
Itisscientificallyuntenabletoassumethatallcandidatesare
eitherauthentictransientsoralldefects.Areasonableworking
assumptionisthatbothpopulationsarepresentinsome
unknownproportion.Fromthisperspective,evenasingle
authenticdetectionamongmanycontaminantswouldvalidate
theeffortandwarrantcontinuedsearch.
Thisdegeneracyisintrinsictoanyattemptatidentifying
NTAsinarchivalmaterial.Twoprimaryexamplesillustrate
thisproblem:
1.NarrowerFWHMsandrounderprofiles.Hambly&
Blair(2024) interpretslightlymoreconcentrated,round
profilesassignsofspuriousdetectionsandmakesan
examplewithVillarroeletal.(2021). However,atmo-
sphericseeingandshort-lived(sub-secondtofew-
second) opticaleventsarealsoexpectedtoproduce
narrowerFWHMsthanlong-exposedstars(Tokovi-
nin2002; Villarroeletal.2025a). Thus,profilesharpness
alonecannotconclusivelydistinguishbetweenartifact
andastrophysicalorigin.Wenote,inpassing,that
Hambly&Blair(2024) havenotattemptedtoapply
theiranalysistothetripletransientreportedbySolano
etal.(2023), whichmighthaveprovidedamore
stringenttestoftheirconclusions.
2.Spatialdistributions.Ahighnumberdensityoftransient-
likefeaturesinoneregioncouldbemistakenfor
evidenceofpoorplatequality.Butthenumberdensity
oftransientsonaplateisnotdiagnostic.IfNTAsexistin
coordinatedswarms,theseswarmscouldspantensof
squaredegrees,easilycoveringentireplates,whileagrid
ofNTAscouldcovertheentiresky.
18
Inambiguousand
uncertaincases—suchastheplateanalyzedinVillarroel
etal.(2021)—additional transientsorartifactsmay
surroundtheninecandidates(seeSupplementaryInfor-
mationofmentionedpaper). Theirpresenceillustrates
thedegeneracyproblem:authenticeventsmaycoexiston
thesameplateasnumerousstar-likedefects,which
makesitessentialtoapplyindependentdiagnosticssuch
asalignmentstatisticsandEarth’sshadowtests.
Becauseoftheambiguityintheseearlycases,weadvocated
formoretargetedsearchesinVillarroeletal.(2022a),
emphasizingparticularlymultipletransientsalignedalonga
line—wherestatisticalanalysiscandecisivelytestwhether
suchconfigurationsoccurbychance.
Moreover,thetemporalcorrelationsbetweenthe1950s
transientsandboththeWashington1952UFOeventsand124
U.S.,Soviet,andBritishnuclearweaponstestsdeserveserious
attention.Evenifindividualeventsremainuncertain,Bruehl&
Villarroel(2025) showsstatisticallysignificantcorrelations
betweensubsetsofthetransientsampleinSolanoetal.(2022)
andhistoricalnuclearactivityandaerialanomalies.Thisalone
contradictstheideathattheentiresampleconsistsofplate
defects.
Finally,oneofthemostrevealingtestsinvolvesEarth’s
shadow.Nomatterhowasymmetricorirregularthedistribu-
tionsofplatedefectsmaybe,theyhavenoplausiblereasonto
avoidtheEarth’sshadow.Incontrast,transientsassociated
withsolarreflectionswould.Thisshadowtestprovidesa
crucialempiricallevertodistinguishbetweenphysical
reflectionsandrandomdefects—andremainsanessentialpart
ofanyvalidationframeworkmovingforward.
Inthispaper,werelyonhypothesistestingacrosslarge
samples—assessingstatisticalcorrelations,spatialalignments,
andEarth-shadowsensitivity—offeringarobustframework
thatremainsvalideveninthepresenceofsubstantialstellar-
likecontamination.Inthefuture,weaimtouseAI-driven
methodstofilterouttransientsthatresembleplatedefectsor
occurinproblematicregionsoftheplates,andtoestablishan
upperlimitonthefractionofobjectsthatmayrepresentNTAs.
Fornow,wewillusethesimplestmethodstosearchfor
candidatesthatshowsignsofsolarreflection.
18
Seee.g.,PatrickJackson’sspherenetwork:https://www.amazon.com/
Sphere-Network-Mr-Patrick-Jackson/dp/B0DXF1RGL6.
3
PublicationsoftheAstronomicalSocietyofthePacific,137:104504(22pp), 2025October Villarroeletal.

4. 3.PredictionsandExpectations
Naturaltransientsoccuratarateseveralordersof
magnitudelowerthanglintsfromartificialobjects.Even
detectingtwonaturaltransientswithinafewarcminutesof
eachotherduringaone-hourexposureisextremelyunlikely.
19
Incontrast,glintscausedbysolarreflectionsfromflat,
highlyreflectivesurfacesathighaltitudes—suchasGSOs—
couldresultinmultiple,simultaneouspoint-liketransients
duringasinglelong-exposureimage.Iftheglintsoriginate
fromthesameobject,theymayappearalignedalonganarrow
bandorstraightline.Insimplegeometries,theglintscouldbe
equidistantandofsimilarbrightness.However,morecomplex
surfacestructuresmayleadtoirregularspacingandvariable
flux(e.g.,Niretal.2021; Villarroeletal.2022a). Alsoobjects
flyinginformationorcoordinatedswarms,mightbefound
alonggeometricpatterns.
Multipletransientsinasingleimagearefrequentlydetected
inmodernautomatedsurveys.Nearlyalltransientswith
durationsshorterthan0.5sarecausedbythisphenomenon,
oftenoriginatingfromsatellitesorspacedebris(e.g.,Corbett
etal.2020; Niretal.2021). Theseeventstypicallyhave
apparentmagnitudesofr∼9–11.Therateofsuchartificial
glintscanreach∼1800eventshr
−1
sky
−1
neartheequator
(Corbettetal.2020), whichwouldoverwhelmanycomparable
phenomenainmodernsurveysunlessspecificallytargeted.
TheredPOSS-Iplates,reachingr∼20magwith∼50minutes
exposures,arestillcapableofdetectingglintsasshortas0.5s,
althoughthefluxisdilutedbyapproximately9mag.
Platedefects,bycontrast,areexpectedtoberandomly
shapedanddistributed.Thechancethatseveraldefects
simultaneouslymimicstar-likepointsourcesandalignalong
anarrowbandissmall.ThemethodproposedinVillarroel
etal.(2022a) identifies“simultaneoustransients”thatappear
withinthesamelong-exposurephotographicplateandare
additionallyalignedwithinanarrowtolerance.Thisalignment
criterionhelpsdistinguishpotentiallyartificialsignalsfrom
randomcelestialorinstrumentalsources.
Forexample,animagewithninetransientsinsidea10×10
arcmin
2
boxmayexhibita4-pointor5-pointalignment,witha
statisticalsignificancebetween2.5σand3.9σdependingon
thegeometry.Forexactprobabilities,wereferthereaderto
Section5inVillarroeletal.(2022a), whichusesthestatistical
frameworkdevelopedbyEdmunds(1981), Edmunds&
George(1985). Even3-pointalignmentsmaybeconsidered
whenthetotalnumberoftransientsinaregionislow.
Alignmentswiththelowestprobabilityofoccurringbychance
shouldbeprioritizedforfurtherexamination,thoughnot
interpretedasconclusiveevidenceofgeosynchronous
reflections.
Takentogether,theseconsiderationsshowthattheoccur-
renceofaligned,simultaneoustransientsonphotographic
platesisanexcellentcandidatesignatureofreflectiveorbital
objects,especiallyintheabsenceofnaturalorinstrumental
explanations.
Whilealignmentsofmultipletransientsprovideastatisti-
callyrobustsignature,itisimportanttonotethatmostglints
causedbysolarreflectionsareexpectedtoappearassingle,
isolatedtransientsonaphotographicplate.Thisfollows
naturallyfromthegeometryofspecularreflection,wherea
glintisonlyvisiblewhentheorientationofarotatingobject
brieflyalignswiththeobserverandtheSun.Assumingalarge
populationofsuchobjectsingeosynchronousorhigherorbits,
themajorityofeventswillnotrepeatandwillappearona
singleplateonly.Thesepoint-likeflashesmaystillexhibit
perfectPSFshapesandaretypicallyabsentinEarth’sshadow,
furtherdistinguishingthemfrombothnaturalandinstrumental
phenomena.Althoughindividualtransientscarrylessstatis-
ticalweight,theoverallrateandbehaviorofsucheventscan
stillbeusedtoidentifyanon-naturalorigin.Asshownin
Villarroeletal.(2022a), statisticalmodelsincorporatingboth
alignedandnon-alignedtransientsoffercomplementaryroutes
fordetectingtechnosignaturesinhistoricaldata.
4.MethodsandSelection
Webaseouranalysisonthecatalogof298,165short-
durationtransientspresentedinSolanoetal.(2022), detected
inredPOSS-Iplateswithtypicalexposuretimesof45–50
minutes.Thesetransientswereidentifiedusinganautomated
pipelinedevelopedaspartoftheVASCOproject.Forfull
detailsonthedetectionmethodology,datacharacteristics,and
vettingsteps,wereferreaderstoSolanoetal.(2022).
Fromthisdataset,wesearchforspatialgroupingsof
transientswithinsquareboxesofvaryingsizes,typically
rangingfromafewarcminutesupto20′–30′ perside(see
typicalsizesinTable2).Foreachgroup,weevaluatewhether
thepositionsofthetransientsfallalongastraightline(ormore
preciselyanarrowband), withinastrometricuncertainties.
WequantifythedegreeofalignmentusingthePearson
correlationcoefficientαbetweenrightascensionanddeclina-
tion.Weretainonlythosecandidatealignmentswhere
|α|>0.99.Wenotethatthecorrelationiscomputedwithout
applyinga()cos
correctiontorightascension.Giventhe
smallangularseparationsinvolved,thishasanegligibleeffect
ontherankingofcandidatealignments.
Table1summarizesthenumberofalignedgroupsfound
withr�3,r�4,r�5,andr�6transients,respectively.
Becausethesearchboxesvaryinsize,thenumberoftransients
pergroupisnotdirectlycomparableacrosscases.
19
Weconsidertheprobabilityofdetectingatransientwithin1hrinthe
POSS-Isurvey,basedonSolanoetal.(2022), whoidentified298,000
transientsover780hrofexposure.Thechanceoffindingonetransientina10
arcmin
2
boxinonehourisapproximately∼0.0016.Theprobabilityoftwo
suchtransientsappearinginthesameboxisthenp∼10
−6
.
4
PublicationsoftheAstronomicalSocietyofthePacific,137:104504(22pp), 2025October Villarroeletal.

5. All83candidatesarepresentedintheAppendix.Visual
inspectionrevealsthatdupletsandtripletsarerelatively
common.However,ratherthanevaluateeveryalignmentwith
N�3,wefocusonhigher-confidencecandidateswithatleast
fouralignedtransients.
ManyPOSS-IplateshavebeenscannedbybothDSSand
theSuperCOSMOSSkySurvey(Hamblyetal.2001). Since
SuperCOSMOSimagesgenerallyofferhigherspatialresolu-
tion,weusedbothsourcestoverifyeachalignment.We
downloadedFITSimagesforallcandidateswithN�4,
selectingimageboxesthatencompassthefullalignment.
Inseveralcases,transientsinitiallyappearingaspoint
sourcesinDSSwererevealed—throughSuperCOSMOS
images—tobeeitherscanningartifactsorrounddefectslikely
causedbyemulsionflaws.Transientsabsentfromthehigher-
resolutionscanwereexcludedfromfurtherconsideration.We
thusretainedonlythosecandidatesthat:
1.Showatleastfourstar-liketransientsinaroughlylinear
arrangementontheDSSscan;
2.AreconfirmedbythecorrespondingSuperCOSMOS
scan.TheDSSandSuperCOSMOSscansareindepen-
dentdigitizationsofthesamephysicalphotographic
plate,obtainedusingdifferentscanners,optics,and
digitizationprocedures.Thismeansthatanyobject
visibleinbothscansisalmostcertainlyarealfeature
presentontheplateemulsion.Incontrast,anobject
visibleinonlyoneofthescansismostlikelyascanning
artifactcausedbydustonthescannerglass,digitization
noise,orcompressioneffects—notagenuineplate
defect.Wethereforetreatagreementbetweenbothscans
asastrongindicatorofauthenticity.Furthermore,we
notethatsomeobjectsthatinitiallyappearpoint-likein
DSSimagesmayexhibitsubtleasymmetriesordeviate
fromastellarPSFinthehigher-resolutionSuper-
COSMOSimages,leadingustorejectthem.This
procedurehelpsensurethattheremainingcandidates
arenotspuriousartifactsintroducedduringdigitization.
Fromthisrefinedset,weidentifyfiveofthemostpromising
candidatesinthenorthernhemisphere,listedinTable2.
Therearetwokeywaysthesearchprocedurecouldbe
improved:
1.Searcharea.ObjectsinGSOmoveat∼10″s
−1
,orabout
10°duringa50minutesexposure.Ourcurrentboxsize
(upto30′)isconservativeandmaymisslonger
alignments.
2.Correlationthreshold.Thecriterion|α|>0.99is
unnecessarilystrictandexcludesmildlycurvedornon-
idealalignments.
However,relaxingeitherparameterwoulddrasticallyincrease
thenumberofcandidates—potentiallyintothetensofthousands
—necessitatingsubstantialmanualvetting.Toaddressthis,we
aredevelopinganexpansionoftheVASCOcitizenscience
platform(Villarroeletal.2022b) tailoredtothistask.
5.TheShortlist
TheshortlistinTable2showsthecandidates.Each
candidateisshowninFigures1–5.Hereweshowonlythe
transientsthemselvestoassistthereader.Thesameimages,
butshowingtheactualalignments,canbefoundin
Figures6–10. Thealignmentsdifferinwidth;therefore,a
dasheddoublelineisshowninsomeparticularcaseswherethe
widthofthestripeislargerthan1″.
Insomecases—forexample,theobjectsmarkedwith
crossesinCandidate3andCandidate5—itisnotcertainthat
everytransientisapointsource,basedoninspectionofthe
images.Slightasymmetriesinthelightprofilesarepresentina
fewcases,manifestingasmildelongations(frome.g.,
movement) orqualitativeirregularitiesinshape.
20
Therefore,
thealignmentispossiblyacombinationoftransientsandplate
defects—orobjectsintheskywithinouratmosphere.The
readercaninspectthehigh-resolutionimagesfromSuper-
Cosmos.
21
Weimprovetheastrometryfortheimagesusingthe
TerapixSWarpprocedure.Wemeasuretheimprovedcoordi-
natesandtheFWHMforeachtransient;seeTable3.Thedates
aretakenfromtheSTScIDSSPlateServer.
Inafewcases,itispossibletoderivemorethanonevariantof
thealignment—forexample,witheithera3-pointora4-point
alignment.Insuchcases,weshowbothoptionsseparatelyinthe
imagesinFigures6–10.Forthecasesintheshortlist,weestimate
theprobabilityofachancealignment;seeSection6.
Table1
Candidates
Region r�3 r�4 r�5 r�6
0<R.A.<100,0<decl.<90 22 5 ⋯ ⋯
100<R.A.<200,0<decl.<90 18 7 ⋯ ⋯
200<R.A.<300,0<decl.<90 32 6 1 ⋯
300<R.A.<360,0<decl.<90 11 2 1 ⋯
Total 83 20 2 0
Note.Totalnumberofalignedtransientcandidatesidentifiedineachsky
region.risthenumberofalignedpoints.Notethatr�4andr�5aresubsets
ofr�3.R.A.anddecl.areindegrees.
20
Theseasymmetriesrefertodeviationsinmorphology,nottothefullwidth
athalfmaximum(FWHM), whichvariesacrosstheplatesduetowell-
documentedinstrumentalandphotochemicaleffects.Asdiscussedin
Villarroeletal.(2025a), thenonlinearresponseofphotographicemulsions
causesbrighterobjectstonaturallyappearwithbroaderprofiles,contributing
totheobservedFWHMspread.
21
http://www-wfau.roe.ac.uk/sss/pixel.html
5
PublicationsoftheAstronomicalSocietyofthePacific,137:104504(22pp), 2025October Villarroeletal.

6. 6.Statistics
Thesectionbelowprovidesabriefrecaptureofthe
statisticalframeworkdevelopedinSection5ofVillarroel
etal.(2022a), whereinterestedreaderscanexplorethedetails
oftheframework.ItbuildsonEdmunds(1981) andEdmunds
&George(1985) whichcriticizedHaltonArp’squasar
alignments.Thecommoncritiquewasthatwithalarge
numberdensityofobjects,alignmentswillinevitablyappear.
Thesepapersdevelopedastatisticalframeworktoinvestigate
theactualprobabilityofchancealignments.
Foreachoftheinterestingcasesweconsiderthetotal
numberNoftransient-likeobjectsfoundintheimagefield,
i.e.,theareaAoftheimage,andlookforrobjectsaligned
withinastripofwidthpmax
andlengthdmax .Suchalignments
willbereferredtoasr-pointalignments.
Figure1.Candidate1.WeshowthecandidateinSuperCosmosscansofPOSS-Ired(left)andPOSS-IIred(right) images(inverted). Transientsaremarkedwith
bluecircles.Thecandidatewithameasuredcoordinateismarkedwithacross(+).Pinkcirclesshowdefects.AlsothegraylinecrossingthePOSS-Ifieldisa
scanningdefect.FourtransientsarevisibleinthePOSS-Iimage,wherethreefollowastraightline.Boxsizeis10×10arcmin
2
.SeeFigure6foraversionwith
drawnlinesthatshowsthepossiblealignment.
Table2
CandidateShortlist
CandidateShortlist
Candidate Year R.A.Decl. R.A.Decl. r N Apmax dmax
μ
r
(sexag.,J2000) (deg,J2000) (arcmin
2
) (arcsec) (arcmin)
1 1954 02:29:33.71+28:31:56.98 37.390445428.5324936 3 4 10×10 1.0 5.8 0.044
2 1955 03:05:42.48+07:58:29.60 46.42698147.9748892 3 5 10×10 1.0 3.6 0.010
3 1954 03:08:27.13+34:40:46.01 47.113023634.6794470 3 5 15×15–16 2.0 9.9 0.194
” ” ”” ” 5
*
5 ” 15.0 ” 0.002
4 1954 21:24:39.71+68:31:30.04 321.165474068.5250111 3 6 12×12 1.0 5.15 0.049
” ⋯ ” ” 4
*
6 ” 5.0 ” 0.003
5 1952 19:16:45.76+51:28:52.40 289.190685451.4812217 3 5 10×10 1.0 4.0 0.028
” ⋯ ” ” 5
*
5 10×10 10.0 4.0 0.0001
Note.Weshowthemostinterestingcandidatesemergingafterthevisualinspection.Insomecasestherecouldbedifferentpossibilitiesofr-pointalignments,e.g.,
r=3orr=4,andweshowbothpossibilitiesmarkedbyanasterisk(
*
).Thegivenpositioncoordinatecorrespondstothetransientmarkedwithacross(+)ineach
figure.
6
PublicationsoftheAstronomicalSocietyofthePacific,137:104504(22pp), 2025October Villarroeletal.

7. AstheareaAisdifferentforeachcase,wecanonlyestimate
theexpectednumberofr-pointalignmentsμ
r
withinagiven
fieldA.AssuggestedinVillarroeletal.(2022a), weusethe
generalizedformulafromEdmunds&George(1985),()
( )
μ= 1
npA
r
xe dx
2
1
,
r
r r
r
d
r xnp
2
max
2
0
12
max
max
Figure2.Candidate2.WeshowthecandidateinSuperCosmosscansofPOSS-Ired(left)andPOSS-IIred(right) images(inverted). Transientsaremarkedwith
bluecircles.Thecandidatewithameasuredcoordinateismarkedwithacross(+).FourtransientsarevisibleinthePOSS-Iimage,wherethreefollowastraightline.
SeeFigure7foraversionwithdrawnlinesthatshowsthepossiblealignment.Boxsizeis10×10arcmin
2
.
Figure3.Candidate3.WeshowthecandidateinSuperCosmosscansofPOSS-Ired(left)andPOSS-IIred(right) images(inverted). Transientsaremarkedwith
bluecircles.Thecandidatewithameasuredcoordinateismarkedwithacross(+)andmightbeslightlydubiousinshape.Pinkcirclesshowdefects,bothplate
defectsandscanningdefects.SeeFigure8foraversionwithdrawnlinesthatshowsthepossiblealignment.Boxsizeisroughly15×15arcmin
2
.
7
PublicationsoftheAstronomicalSocietyofthePacific,137:104504(22pp), 2025October Villarroeletal.

8. Figure4.Candidate4.WeshowthecandidateinSuperCosmosscansofPOSS-Ired(left)andPOSS-IIred(right) images(inverted). Transientsaremarkedwith
bluecircles.Thecandidatewithameasuredcoordinateismarkedwithacross(+).SeeFigure9foraversionwithdrawnlinesthatshowsthepossiblealignment.
Boxsizeis12×12arcmin
2
.
Figure5.Candidate5.WeshowthecandidateinSuperCosmosscansofPOSS-Ired(left)andPOSS-IIred(right) images(inverted). Transientsaremarkedwith
bluecircles.Thecandidatewithameasuredcoordinateismarkedwithacross(+).SeeFigure10foraversionwithdrawnlinesthatshowsthepossiblealignment.
Boxsizeis10×10arcmin
2
.
8
PublicationsoftheAstronomicalSocietyofthePacific,137:104504(22pp), 2025October Villarroeletal.

9. whereΓisthegammafunction,n=N/A, andallother
quantitiesareaspreviouslydefined,withlengthsgivenin
arcminand,consequently,theareaAisinarcmin
2
.Asin
Villarroeletal.(2022a) weuse,forpracticalreasons,a
limitingcaseofthisgeneralization,( )!
()
μ =
npdA
rr
r
2
2
, 3,4,5,, 2
r
r r
r r
2
max
2
max
Figure6.Candidate1.WeshowthecandidateinSuperCosmosscansofPOSS-Ired(left)andPOSS-IIred(right) images.Transientsaremarkedwithgreencircles.
Thecandidatewithameasuredcoordinateismarkedwithacross(+).Adashedwhitelineshowsthealignment.Yellowcirclesshowdefects.Alsothewhiteline
crossingthePOSS-Ifieldisascanningdefect.Wesee4transientsinthePOSS-Iimageswherethreefollowastraightline.
Figure7.Candidate2.WeshowthecandidateinSuperCosmosscansofPOSS-Ired(left)andPOSS-IIred(right) images.Transientsaremarkedwithgreencircles.
Thecandidatewithameasuredcoordinateismarkedwithacross(+).Adashedwhitelineshowsthealignment.Yellowcirclesshowdefects.Alsothewhiteline
crossingthePOSS-Ifieldisascanningdefect.Wesee4transientsinthePOSS-Iimageswherethreefollowastraightline.
9
PublicationsoftheAstronomicalSocietyofthePacific,137:104504(22pp), 2025October Villarroeletal.

10. whichisagoodapproximationwhendpn2 1
max max and
simplifiesthecalculationsconsiderably.Forthepresentstudy
Equations(1)and(2)shouldyieldverysimilarresults,sincedpn2 0.01
max max
forallcasesconsidered.
WeapplyEquation(2)tocalculatetheexpectednumberof
r-pointalignmentsμ
r
foreachcase.Weincludeallmeasure-
mentsinTable2.Theshortlistincludesboth3-point
alignmentsand4-pointalignments.Sinceeachcandidatecase
onlyhasonealignment,theprobabilityisgivenbythe
expectationvalue,μP r
.Wecanseethatseveralofthecases
aresignificantlystatisticallyimprobable(3σ–4σ) tohappenin
asingleimage.
Theprobabilityestimateisalsoverysensitivetothetotal
numberoftransients(N)present.Thisnumberdepends
Figure8.Candidate3.WeshowthecandidateinSuperCosmosscansofPOSS-Ired(left)andPOSS-IIred(right) images.Theupperrowshowsa3-pointalignment
within1″–2″. Thelowerrowshowsa5-pointalignmentofwithin15″.Transientsaremarkedwithgreencircles.Thecandidatewithameasuredcoordinateismarked
withacross(+)andmightbeslightlydubiousinshape.Thedashedlinesshowsthealignment(thewhitedoublelineforthethickeralignmentbelow). Yellowcircles
showdefects,bothplatedefectsandscanningdefects.
10
PublicationsoftheAstronomicalSocietyofthePacific,137:104504(22pp), 2025October Villarroeletal.

11. stronglyonthevisualinspectionthatwasmadebyblinkingthe
POSS-IandPOSS-IIimagesinSAOImageDS9,takinginto
accountthedifferencesindepth.Anymissedtransientswill
changethevalueofN,andhencetheestimatedprobability.
However,toestimateexactlythetotalprobabilityPtot
of
eachsingleeventtohappenduringoursearches,twomore
factorsinfluencethetotalprobability.Thefirstisthe
probabilityforobtainingaperfectly1,2,…orN
star-like
plateflawswithinthesameareaofanimage.Giventherarity
ofencounteringastar-likeplatedefect,andevenlesssowitha
matchingFWHMasthenormalstarsofthesamemagnitude
rangeinthefield,itmaybeevenmoreunusualtoencounter2,
or3,or4plateflawsthatallhavethesamecoincidental
featuresandthislowersdramaticallythetotalprobabilityof
theevent.Thesecondfactoristhetotalnumberofmultiple
transientsinourdataset:iftherearesufficientlymanystar-like
plateflawscausing“multipletransients,”someofthesewill
lineup.Withaninfinitedataset,anytypeofconstellationswill
befound.Thisfactorwill,contrarytothefirstfactor,increase
thetotalprobabilityforaneventtooccur.
Unfortunately,wehavenograspormeansofestimating
eitherofthetwofactors.Therefore,itiseasiertoexaminethe
Figure9.Candidate4.WeshowthecandidateinSuperCosmosscansofPOSS-Ired(left)andPOSS-IIred(right) images.Theupperrowshowsa3-pointalignment
within1″.Thelowerrowshowsa4-pointalignmentofwithin5″.Transientsaremarkedwithgreencircles.Thecandidatewithameasuredcoordinateismarkedwith
across(+).Thedashedlinesshowsthealignment(thewhitedoublelineforthethickeralignmentbelow).
11
PublicationsoftheAstronomicalSocietyofthePacific,137:104504(22pp), 2025October Villarroeletal.

12. effectofthechoiceofpmax ontheprobabilityestimatesfor
singleimages.Thechoiceofpmax
dependsonthescience
questionofinterest:areweinterestedwhethertheobjectsare
trulyalignedorwhethertheyarejustnon-random?Showing
non-randomnessisallthatisneededtoargueforthe
authenticityofthepoints,butnotnecessarilyenoughtoargue
thattheytrulyarealignedasinthecaseofGSOglints.Weuse
Table3toadoptothervaluesofpmax
,settingitequalto
FWHMofthesmalleststarinanalignment(e.g.,forCandidate
1,FWHM=2.
7).Doingthis,weseethatall3-point
alignmentsarenon-interestingeventswithp>0.05(less
significantthan2σ),withanexceptionoftheborderlinecase
ofCandidate2.ThisshowsthatforPOSS-Idatawherethe
seeingingeneralisratherlarge,3-pointalignmentsof
simultaneoustransientsdonotprovidesignificantproof
againstrandomness.Theinterestingcasesarethe4-pointand
Figure10.Candidate5.WeshowthecandidateinSuperCosmosscansofPOSS-Ired(left)andPOSS-IIred(right) images.Theupperrowshowsa3-point
alignmentwithin1″.Thelowerrowshowsa5-pointalignmentofwithin10″.Transientsaremarkedwithgreencircles.Thecandidatewithameasuredcoordinateis
markedwithacross(+).Thedashedlinesshowsthealignment(thewhitedoublelineforthethickeralignmentbelow).
12
PublicationsoftheAstronomicalSocietyofthePacific,137:104504(22pp), 2025October Villarroeletal.

13. 5-pointcases,namelyCandidates3,4and5.Yet,onecould
arguethatwithoutaninspectionwithamicroscopeonestill
cannotexcludeplatedefects.
However,whatmakestheeventsevenmoreinterestingis
thatCandidate5occursonthesamedateasoneofthemost
famousUFOmasssightingsinhistory—namely,the1952
WashingtonUFOflap(Villarroel2024). Thiscouldbea
coincidence.WealsonotethatCandidate1occurswithinaday
ofthepeakofthe1954UFOwave.Weshalldiscussthis
furtherinSection10.Theseadditionaltwocoincidences
furthermotivatescrutinyoftheplatedefecthypothesis,
especiallyinlightofthecombinedstatisticalandcontextual
factorspresentedinthisstudy.
7.AssessmentofConventionalExplanations
Thecentralchallengeofthisworkliesindetermining
whetherthetransientsrepresentauthenticobservations.A
previousanalysisofthemultipletransienteventinVillarroel
etal.(2021) ruledoutallknownastrophysicalorigins,and
mostinstrumentalcausesaswell.Whatremainsisthe
possibilityofunknownplatecontaminationoremulsion
defectsthatcoincidentallyresemblestar-likeshapes,despite
theirvariationinbrightness.Whilegravitationallensingbya
short-livedtransientpassingbehindanundetectedsuper-
massiveblackhole(SMBH) wasproposedinSolanoetal.
(2023), suchamodelwouldrequireanimplausiblylarge
populationofundetectedSMBHsintheMilkyWaytoexplain
thebroadersetofeventsfoundbyVASCO.Thephenomenon
remainsunresolved—nowmadeevenmoreintriguingbythe
discoverythatseveraleventsarealignedalonganarrowband.
Apotentialconcernisopticalghosts.Ghoststypically
exhibitextendedorclumpymorphologiesanddonotmatch
stellarpointspreadfunctions(PSFs). Incontrast,thetransients
identifiedinbothVillarroeletal.(2021) andthecurrentstudy
werepreselectedbasedontheirPSF-likeproperties(Solano
etal.2022), makingclassicalghostinganunlikelyexplanation.
WhilemostopticalghostsonPOSS-Iplatesappearextended
orirregular,onemightaskwhethermoreunusualghosting
patterns—suchaspoint-likereflections—couldinprinciple
occur.ModernCCD-basedsurveysusingthesametelescope
(e.g.,ZTF,PTF) havedocumentedrareghostpatternsthat,
underspecificopticalconditions,canmimicpointsourcesat
significantangularseparationsfromtheirparentstars(Waszc-
zaketal.2017; Irureta-Goyenaetal.2025). Adedicated
opticalmodelingeffortwouldberequiredtofullyevaluate
suchhypotheticalscenarios,takingintoaccountthedifferent
detectorareas(CCDsversusphotographicplatefieldsizes) and
pixelscales.However,suchananalysisliesbeyondthescope
ofthepresentstudy.Instead,wecomparetwoimagesofthe
samefieldtakenwiththesameconfiguration,separatedby
only30minutes:theredandtheblueplatesforthefive
Table3
Measurements
Candidates1–5
Object R.A. Decl. FWHM FWHM R
(sexag.,J2000) (pixel) (arcsec)
object1 2:29:37.57+28:36:31.58 4.0 2.7 18.9
object2
*
2:29:21.38+28:36:57.89 7.2 4.8 16.6
object3
*
2:29:21.76+28:36:49.09 7.6 5.1 17.0
object4†
*
2:29:33.80+28:31:56.83 4.1 2.7 18.3
Dateofobservation=1954-10-04
Object R.A. Decl.(sexag.,
J2000)
FWHM
(pixel)
FWHM
(arcsec)
R
object1 3:05:52.34+8:00:16.97 3.8 2.5 19.2
object2†
*
3:05:42.46+7:58:30.22 10.0 5.7 15.2
object3
*
3:05:42.81+7:58:20.56 5.9 4.0 17.9
object4
*
3:05:50.24+7:55:33.86 4.4 2.9 18.3
Dateofobservation=1955-01-14
Object R.A. Decl.(sexag.,
J2000)
FWHM
(pixel)
FWHM
(arcsec)
R
object1
*
3:08:29.90+34:31:25.73 6.2 4.2 17.1
object2
*
3:08:30.72+34:31:27.44 5.2 3.5 18.1
object3†
*
3:08:27.42+34:40:46.00 9.9 6.6 15.4
object4
*
3:08:27.05+34:41:13.49 8.1 5.4 16.1
object5
*
3:08:26.56+34:41:07.89 6.0 4.0 17.1
Dateofobservation=1954-12-21
Object R.A. Decl.(sexag.,
J2000)
FWHM
(pixel)
FWHM
(arcsec)
R
object1 21:24:45.51+68:34:00.29 4.4 2.9 N.A.
object2 21:24:44.59+68:34:01.20 4.6 3.1 16.6
object3
*
21:24:47.62+68:31:58.92 4.4 2.9 17.9
object4†
*
21:24:39.72+68:31:31.22 8.9 6.0 15.2
object5
*
21:24:38.18+68:31:27.97 5.0 3.4 17.2
object6
*
21:24:03.94+68:29:14.36 4.6 3.1 17.8
Dateofobservation=1954-08-05
Object R.A. Decl.(sexag.,
J2000)
FWHM
(pixel)
FWHM
(arcsec)
R
object1
*
19:16:51.46+51:30:24.51 11.0 7.4 13.2
object2
*
19:16:50.64+51:30:20.86 12.0 8.0 12.7
object3†
*
19:16:45.73+51:28:52.04 7.2 4.8 16.0
object4
*
19:16:40.13+51:27:12.85 5.0 3.4 16.3
object5
*
19:16:40.27+51:27:06.29 5.5 3.7 16.0
Dateofobservation=1952-07-27
Note.Welisttheastrometry-improvedmeasurementsfortheobjectsinsidethe
greencirclesinFigures1–5.Objectsthatareplacedinsideanalignmentare
markedwithanasteriskm∗.ThecentralobjectspresentedinTable2are
markedwithadagger(†).WeshowtheFWHMinpixelandarcsec,basedon
SuperCosmosPOSS-Iimages.TheSuperCosmosresolutionis0.
67pixel
−1
.
TheobjecthaveanimprovedastrometrywithhelpofTerapixswarp
procedure,usingzero-pointcalculationswithSDSSasareferencefield.The
rmagnitudesareobtainedviathephotometricproceduredescribedby
B.Villarroeletal.(2025inpreparation) forDSSscannedPOSS-Iredplates,
buildingonmethodsbyAndruketal.(1995, 2017, 2019). Whenanobject
eitheristoofaintortwoobjectsaretooclosetoeachother,thephotometry
code(thatmeasuresRJohnsonmagnitudes) failstodetectthem,meaningwe
havenophotometricinformation.Forthesecases,wemarkthemagnitudesas
NonAvailable(N.A.).
13
PublicationsoftheAstronomicalSocietyofthePacific,137:104504(22pp), 2025October Villarroeletal.

14. candidates.Thisissimilartothecaseofthetripletransient,
wherestarsofabout16thmagnitudeappearandvanishwithin
halfanhour,yetarenotseenonthecorrespondingblueplate
withsimilarconfiguration.Theblueexposuresaresignifi-
cantlyshorter,yetwedonotseeanysimilarstarsorrepeating
patternsinthem,seeforexampleFigure11.Thisfurther
reducesthelikelihoodthatopticalghostingcanexplainthe
alignments,thoughdifferencesinexposuretimeandinthe
reflectivitiesofopticalcoatings(causingdifferentghosting
levelsindifferentcolorbands) preventsusfromfully
excludingit.
Anotherconcernisphotographicplatedefects.Historically,
astronomershaveexcludedsingle-epochpointsourcestoavoid
falsepositives—anapproachthatalsoinadvertentlyeliminate
manygenuine,short-livedastronomicalevents.Forexample,
Hambly&Blair(2024) arguedthatthetransientsreportedin
Villarroeletal.(2021), despitetheirpoint-likemorphology,
arelikelyemulsionartifacts.Thisconclusionwasprimarily
basedonthefindingthatthetransientsexhibitslightly
narrowerfullwidthathalfmaximum(FWHM) values,on
average,comparedtonormalstars.However,theanalysisdid
notaccountfortheknownnonlinearityofphotographic
emulsions,whichcausesfaintersourcestonaturallyexhibit
narrowerprofiles.Inaddition,the“artifact”sampleintheir
studywasselectedusingcriteriathatmirrortheVASCO
project’stransientselectionpipeline,whichmayintroduce
circularreasoning.Crucially,thestudydidnotconsiderthat
sub-secondopticalflashesarepredicted—onphysicalgrounds
—toappearsharperandmorecircularthanstarsinlong-
exposureplates,duetotheabsenceofatmosphericseeing,
windshake,andtracking-inducedsmearing.Theseeffectsare
discussedindetailinadedicatedtechnicalcommentary
(Villarroeletal.2025a). Todate,nostudyhassystematically
quantifiedthefractionofsingle-platedetectionsthatare
authentictransientphenomenaversuscoincidentallystar-like
emulsiondefects.
Asummaryofexcludedastrophysical,observational,and
instrumentalcausesisprovidedinVillarroeletal.(2021).
Assumingtheobservedtransientsaregenuineandnotartifacts,
weturntoalternativephysicalexplanationsbeyondtheGSO
glinthypothesis.
Point-likeeventscouldresultfromeitherreflectedsunlight
orintrinsicemission.AsshowninVillarroeletal.(2021), such
objectsmustbelocatedwithinthesolarsystem.Weconsider
fourbroadpossibilities:(i)theobjectsareinsideEarth’s
atmosphere,(ii)theyareinlowEarthorbit(LEO), (iii)they
areinGSO,or(iv)theyarelocatedatsignificantlygreater
distances.
Ifthetransientsoriginatedfromluminousorreflective
atmosphericobjects,theyshouldleavevisibletrailsoverthe
45–50minutesPOSS-Iexposures,giventhatthetelescope
trackedstarsduringimaging.Stationaryobjectswouldalso
appearstreaked.Objectsveryclosetotheobserverwould
appearsignificantlyoutoffocusduetoproximitytothefocal
plane.Forinstance,anobjectat50kmaltitudewouldsuffera
defocusofseveralhundredmicronsonthePalomar48inch
system,resultinginanextendedPSFincompatiblewitha
stellarappearance.Onlyataltitudesaboveseveralhundred
kilometerswouldpoint-likemorphologybeachievable.This
effectivelyrulesoutphenomenasuchasredspritesorrare
Figure11.Redvs.blueimageofCandidate5.Comparisonbetweenthered(left)andblue(right) DSSimagesforCandidate5.Nocoincidentsourcesarefoundin
theblueimage.
14
PublicationsoftheAstronomicalSocietyofthePacific,137:104504(22pp), 2025October Villarroeletal.

15. luminousatmosphericeventsliketheHessdalenphenomenon
(Teodorani2004). Theonlyplausiblescenarioinwhich
multipleobjectswithinEarth’satmospherecouldproduce
point-liketransientswithoutvisibletrailsisiftheywerelight-
emittingandappearedsimultaneouslyforasplitsecond—brief
enoughtoavoidleavingmotionblur—beforevanishing.
However,suchobjectswithintheatmospherewouldbesubject
tofocalblur.Alternatively,theywouldneedtomimicthe
appearanceofstarsasseenfromEarth.Whilespeculative,
suchascenariocannotberuledoutaprioriandwouldfall
underthecategoryofunidentifiedaerialphenomena.Some
asymmetriesobservedin,forinstance,Candidate5,mightstill
bemarginallyconsistentwithhigh-altitudesourcesnearthe
upperatmosphere.Allplausiblescenarioswouldfitwiththe
observationsofUAP,seeKnuthetal.(2025).
LEO-basedexplanationsarenotimpossible,buttheyare
muchlesslikely.PSF-likeglintsduetoshortmillisecond
flashescanbeproducedatanyorbitaltitudebyrapidly
spinningobjects.Nevertheless,objectsinLEOtypicallyleave
continuoustrails,andexplanationsinvolvingglintsfrom
experimentalrocketsormissilesataltitudesof100–200km
areimprobableduetotheirrapidmotionandconstrained
illuminationgeometry.Further,empiricalstudiesbasedon
short-exposureCCDsurveys(e.g.,Corbettetal.2020; Nir
etal.2021) haveshownthatmostPSF-likeglintsare
associatedwithGSO.However,ifanobjectwerecapableof
activelycontrollingbothitsmotionanditsopticalsignatureas
perceivedfromEarth-basedobservatories,thenaltitude
constraintswouldnolongerapply.Suchascenariowould
implyanengineeredsystemofextraordinarysophistication.
Wealsoconsideredmoredistantorigins.Asdiscussedin
Villarroeletal.(2021), fast-movingSolarSystemobjectssuch
asasteroidswillproducetrails,whileslow-movingones
shouldappearinmultipleimagestakencloseintime.Objects
liketumblinginterstellarbodies(e.g.,‘Oumuamua) wouldalso
producevisibletrailsacrosslongexposures.Hence,noknown
populationofsolarsystemorinterstellarobjectscanexplain
point-liketransientsthatappearonlyinonelongexposureand
areentirelyabsentshortlybeforeandafter.
Whilewecannotexhaustivelyruleoutallpossible
explanations,includingthosenotyetimagined,theabsence
ofknownnaturalorinstrumentalcauses—combinedwiththe
spatialalignmentofcertaineventsalonganarrowband—calls
forfurtherinvestigation.Andmaybethesimplestwayof
testingthemechanismbehindtheseflashes,isbyperforminga
testthatcanrevealwhethertheyoriginatefromsolar
reflections—orifnot.
8.TestingtheSolarReflectionHypothesis
TheVASCOprojecthasidentifiedthousandsofshort-lived,
point-liketransientsinpre-Sputnikphotographicplates
(Villarroeletal.2020; Solanoetal.2022). Themultiple
transientcandidateswerefoundamongthisgeneralpopula-
tion,withseveraleventssharingsimilartimescales,morphol-
ogies,andapparentmagnitudes.Itisthereforereasonableto
treatthemultipletransientsasastatisticallyidentifiable
subpopulationwithinthisbroaderdistribution.
Onepossibleinterpretationfortransientsisthattheyare
causedbysunlightreflectingoffobjectswithflatsurfacesin
GSOs,suchassmallrotatingobjectsbrieflyglintingasthey
passthroughafavorableviewinggeometry(Villarroeletal.
2022a). Ifthisinterpretationholds,wewouldexpecta
significantdeficitofsucheventswithinEarth’sshadow
(umbra), wheresunlightcannotreachtheobjecttoproducea
glint.Ifthetransients,ontheotherhand,arecausedbytheir
ownemissionorareduetoplatedefects,wewouldexpectno
deficitinthenumberoftransientswithintheshadow.The
methodofusingEarth’sshadowtofilteroutreflectionsis
furtherdiscussedinVillarroeletal.(2025b).
Whileitispossibletocomputethefractionofeach
photographicplatethatliesinEarth’sshadowforanygiven
orbitalaltitude,notallheightsareequallymeaningfulforour
analysis.Atlowaltitudes(e.g.,below∼10,000km),Earth’s
shadowmaycoverlargefractionsoftheplates,makingany
deficitorsurplushardtointerpret.Whileplatedefectsdonot
respondtothepositionofEarth’sshadow,thediagnostic
powerofthistestdependsontheassumptionthattheshadow
israndomlyplacedwithrespecttoplategeometryandartifact
distribution.Whentheshadowcoversalargeportionofthe
plate(e.g.,>50%), thisassumptionbreaksdown,andevena
randomdistributionofartifactswillnaturallyyieldan
overdensityintheshadowedregion.Insuchcases,thetest
becomeslesssensitivetosystematicavoidance,makingsmall
shadowcoverages(e.g.,<5%–10%) morereliable.
Moreover,reflectiveobjectsinloworbitstendtomove
rapidlyandwouldoftenappearasstreaksratherthanpoint-
spread-function(PSF)-like transients.Sinceoursampleonly
includesPSF-likedetections,itisphysicallyunlikelythat
manyofthemoriginatefromlowEarthorbits,whereglints
wouldneedtobeextremelyshort-lived(ontheorderof
milliseconds). Niretal.(2021) showthatmostsub-second
flaresareglintsofsunlightreflectedfromsatellitesin
geosynchronousandgraveyardorbits.Forthesereasons,we
focusourmainanalysisonaltitudeswherelessthan5%ofthe
fieldistypicallyshadowed—regionswheretheshadow
behavesapproximatelyrandomly,andwherereflectiveglints,
ifpresent,wouldbebothdetectableandphysicallyplausible.
WeusethetransientcandidatesfromSolanoetal.(2022),
butwiththeadditionalrequirementthattheyhaveno
counterpartswithin5″inGaia,Pan-STARRSandNeoWise.
Furthermore,werestrictouranalysistoobjectsinthenorthern
hemisphere(decl.>0°).Thisyieldsasampleof106,339
transients,whichweuseforourstudy.
Animportantnoteaboutthesampleisthat,contrarytothe
othertransientcandidatesdiscussedthroughoutthepaper,this
15
PublicationsoftheAstronomicalSocietyofthePacific,137:104504(22pp), 2025October Villarroeletal.

16. samplehasnotbeenvisuallyinspected.Assuch,itisexpected
tocontainasubstantialnumberoffalsepositives,including
clusteredartifactssuchasedgefingerprintsorotherplate
defectsthatcontaminateoursample.Inaddition,thespatial
distributionofthesampleisnotisotropicduetoinhomoge-
neousskycoverageintheoriginalPOSS-1survey.Some
regionsoftheskyaremoredenselysampledthanothers,
leadingtovariationintheoveralldetectiondensity.
However,theseeffectsdonotbiastheresultsofourshadow
analysis.Thereasonisthatwearecomparingasmall,well-
definedsubsetofthispopulation—thosethatfallwithinthe
Earth’sshadowconeatthetimeofobservation—withtherest
ofthesamepopulation.Sincetheselectioneffectsand
potentialfalsepositivesaffectboththeshadowedand
unshadowedregionssimilarly,anylargeandstatistically
significantdifferenceindetectionratesbetweentheseregions
mustreflectanintrinsicpropertyofthedetectionsthemselves,
notanobservationalbias.Ortoexpressitsimply:platedefects
donotknowwheretheEarth’sshadowis,andhavenoreason
toavoidthatregionmorethananyother.
Thefractionoftransientsexpectedwithintheumbra
dependsontheangularradiusofEarth’sshadowatdifferent
altitudes.Weusethesoftwarelibraryearthshadow
publishedbyGuyNirpublished(GuyNir’scode2024) to
estimatethesizeoftheEarth’sShadowat40,000km(8°.
69)
and80,000km(4°.
57).Thecodedetermineswhetheragiven
pointataspecifiedaltitudeandgeographicpositionliesinside
Earth’sshadow,basedonthesolarangleandthegeometric
configurationoftheSun,Earth,andtheobject.Weapplyitto
eachtransientusingtheirJ2000coordinatesandJulianDates.
Wecomparetheexpectedandobservedratesfortwodifferent
altitudescapableofproducingPSF-liketransients,namely
42,164kmand80,000km.Wecancalculatetheexpectations
basedonhowlargefractionofthenorthernhemisphereis
coveredbytheshadow,andcomparewiththeobserved
fractions.Wecalculatetheareaintwodifferentways,both
basedonbasedonsphericalgeometry:(– )21cos
)aswellas
planarskycoverage,asanapproximation.Table4showsthe
results.Wenotethatmultipleplates(∼10) fallwithinEarth’s
shadow,sotheobserveddeficitisnotdrivenbyasingleoutlier
plate.
Toestimatethestatisticalsignificanceofthedifferencein
transientdetectionrateswithinEarth’sumbraatdifferent
altitudes,wecomputePoissonuncertaintiesfortheobserved
andexpectedfractions.At42,164kmaltitude,weexpect
N=1223transientsinshadowoutof106,339total,corresp-
ondingtoanexpectedfractionof= ±f0.0115
0.00033
exp
.
However,weobserveonlyN=349transientsinshadow,
yieldingf
obs
=0.00328±0.00018.Thedifferencebetween
thesefractionsishighlysignificant,withasignificancelevelof
21.9σ, computedbycombiningthePoissonuncertaintiesin
quadrature:( )( )
=
+
=
+
ff
0.01150.0
0328
0.00033 0.
00018
21.9.
expobs
exp
2
obs
2
2 2
At80,000kmaltitude,weexpectN=339transientsin
shadowoutof106,339total,correspondingtoafractionof= ±f0.0031
90.00017
exp
.However,wefindonlyN=79
transientsinshadow,yieldingf
obs
=0.00074±0.000084.The
differenceintheseobservedfractionsisalsohighlysignificant,
withasignificancelevelof12.7σ, computedbycombiningthe
Poissonuncertaintiesinquadrature:( )( )
=
+
=
+
ff
0.003190.
00074
0.00017 0.
000084
12.7.
expobs
exp
2
obs
2
2 2
Thisresultfurtherstrengthenstheconclusionthatsunlightis
necessaryforproducingthetransientevents.Thestrongdeficit
oftransientswithintheEarth’sumbrasuggeststhatthe
majorityoftheseeventsdependonsunlightillumination,
consistentwiththeglinthypothesis.Thisstronglydefiesthe
platedefecthypothesisandmanyofthealternativehypotheses
presentedinSection7.
Weperformedanadditionaltesttoestimatetheactual
fractionofthesurveyskythatwascoveredbyEarth’sshadow
duringtheactualPOSS-Iobservations,andtocompareittothe
Table4
ComparisonofEarth’sUmbralShadowCoveragewithObservedTransientFractionsintheNorthernCelestialHemisphere(20,626.5squaredegrees)
Alt. θ N A
sph
A
pl
f
sph
f
pl
f
obs
f
sph
/f
obs
(km) (deg)
42,164 8.69 349 237.4 237.2 0.0115 0.0115 0.00328 3.50
80,000 4.57 79 66.0 65.6 0.0032 0.0032 0.00074 4.32
Note.Weshowthealtitude(km),theshadowradiusindegrees(θ),thenumberNofVASCOtransientsdetectedinsidetheshadow,theshadowareaA
sph
assuming
sphericalskygeometry(sq.deg),shadowareaA
pl
assumingplanarapproximation(sq.deg),expectedfractionf
sph
oftransientsinshadowusingsphericalarea,
expectedfractionf
pl
usingplanararea,theobservedfractionf
obs
ofVASCOtransientsinshadow,andtheratiof
sph
/f
obs
.
16
PublicationsoftheAstronomicalSocietyofthePacific,137:104504(22pp), 2025October Villarroeletal.

17. actualobservedfractionoftransientsfallingwithinthe
shadow.Thetransientsampleisbasedon635unique
photographicplates,eachwithadesignatedcentralcoordinate
(R.A.,decl.) inJ2000andacorrespondingobservationtime.
Eachplatespans6°×6°onthesky,aslistedbySTScI.We
simulated180randompointsperplate,foratotalof114,300
points.Foreachsimulatedpoint,wetestedwhetheritwould
fallwithinEarth’sshadowatageosynchronousaltitude
(42,164km)duringa50minutesexposurestartingfromthe
recordedobservation.
Outofthe114,300simulatedpoints(180pointsperplate),
610werefoundtoliewithinEarth’sshadow,implying
thatapproximately0.53%ofthesurveyareashouldbe
shadowedatGSO.However,inouractualtransientdataset,
only349/107,875, 0.32%oftheeventsoccurwithinthe
shadow,correspondingtoa∼39%deficit,significantat
the7.6σlevel.Werepeatedthesameprocedureatahigher
altitudeof80,000km.Inthiscase,theactualshadowcoverage
dropsto(109/114,300) 0.1%,whiletheobservedfractionof
transients(76/107,875) withintheshadowisonly0.07%—a
∼26%deficit,significantatthe2σlevel.Thesimulatedpoints
withineachplateareusedsolelytoestimatetheexpected
geometriccoverageoftheEarth’sshadowduringtheexposure
time,andarenotmeanttorepresentthespatialdistributionof
actualtransients.Thisalsosuggeststhatalargerfractionof
objectsmaybelocatednearGSOthanat80,000km,although
thelimitednumberofeventsat80,000kmmakesthe
comparisonstatisticallyuncertain.
Weperformanadditional,conservativetestonthe
transients,thistimeassumingatotalexposuretimeof
50minutes.Whileourmainshadowtestassumesthatthe
transienteventoccursatasinglemoment(whichisreasonable
giventheirshortduration), wenowtestwhethertheEarth’s
shadowpassesthroughthetransient’spositionatanytime
duringa50minuteswindow.Thisincreasesthechancethatthe
transientwouldfallwithintheshadow.Wefindthat387
(0.3587%) areintheEarthShadowat42,168km,and80
(0.072%) at80,000km.Evenunderthisgenerousupperlimit
assumption,whereatransientisconsideredshadowedifthe
Earth’sshadowpassesthroughitspositionatanyduringa
50minutesexposure,thedeficitremainsstrong.Thisresult
providesrobustevidencethattheVASCOtransientssystem-
aticallyavoidEarth’sshadow,consistentwithapopulationof
reflectiveobjectsthatareonlyvisiblewhilesunlit.
Thenormalizationtechniquepresentedhereisgroundedina
directsimulationofshadowcoveragebasedontheactual
photographicplatesusedinthesurvey.Eachplate’sposition
andobservationwereusedtosimulateuniformlydistributed
testpointsacrosstheplatearea,allowingustoempirically
estimatetheexpectedfractionofthesurveyskythatfalls
withinEarth’sshadow.Thisapproachminimizesassumptions
andavoidspotentialsystematicbiasesthatmayarisefrom
analyticalsolidangleapproximations.Toavoidintroducing
spatialselectionbias,weincludeallobservedtransientsinthe
analysis,includingthoseclusteredneartheedgesoftheplates,
sinceplatedefectsdonotknowwhereistheEarth’sshadow.
Asaquickcheck,nevertheless,wealsotestbymaskingedge
transients(>2°fromplatecenter) toremoveallartifactsclose
totheplateedge.Removingtheedgeoftheplateinthe
analysis,yieldsasimilar∼30%deficitinEarth’sshadow,
thoughwithborderlinesignificance(2.6σ) duetothedecrease
insamplesizes.
Asanote,atlowaltitudes—wheretheshadowcoversa
largefractionoftheplate—itisalsopossibletoobservea
significantoverdensityoftransientsintheshadowedregion.
Thisisanaturalconsequenceofthegeometriccoverage:when
mostofthefieldliesinshadow,anytransient—regardlessof
origin—isstatisticallymorelikelytofallthere.Suchover-
densitiesarethereforenotphysicallymeaningfulandcannotbe
usedtoinferthenatureoraltitudeoftheobjects.Wetherefore
recommendrestrictingtheanalysistoaltitudeswherethe
Earth’sshadowcoversnomorethan5%oftheplatearea,in
ordertopreservetheassumptionthatitsplacementis
effectivelyrandomwithrespecttoplategeometryanddefect
distribution.
Animportantimplicationofthisanalysisisthatthetotal
numberofglintingobjectsnearGSOmaybesignificantly
underestimatedifoneonlyconsidersthealignedtransient
candidates,sincetheyrepresentonlyaminorsubset(albeit
visuallyvetted) ofthefulltransientpopulation.Ourresults
suggestthatamuchlargerpopulationofobjectscapableof
producingsunlightreflectionsexists,asinferredfromthefull
VASCOtransientsample.Thissystematicdeficitoftransients
inEarth’sshadow—especiallyataltitudeswheresunlight
reflectionsdominate—supportstheinterpretationthata
significantfraction,roughly∼1/3rdofallVASCOtransients,
arecausedbyhighlyreflectiveobjectsinGSO.However,in
ordertodeterminetheabsolutenumberofsuchobjects,we
wouldneedtoquantifythetruefractionoffalsepositivesin
thesample—suchasartifactsandplatedefects—amajor
undertakingthatwillbeaddressedinaforthcomingstudy.
9.TheGSOHypothesis
9.1.ObjectProperties
Inthissection,wediscusstheconditionsunderwhich
reflectionsfromobjectsinGSOcouldproducetheobserved
glints.
Animportantquestioniswhattypesofobjectshapesand
reflectivegeometriesarecapableofcreatingthetransient
signaturesobservedinthePOSS-Iplates.Arapidlyspinning
objectmayproducemultipleglintsduringa50minutes
exposure,whereasamoreslowlyrotatingobjectmight
generateonlyoneortwo.
Ifweassumeafastspinratedinterprettheobservedstripe
lengthd
max
ascorrespondingtothepathtraversedbythe
17
PublicationsoftheAstronomicalSocietyofthePacific,137:104504(22pp), 2025October Villarroeletal.

18. objectduringtheexposure,wecanestimateaprojected
velocityofapproximately0.
5s
−1
.Thisissignificantlyslower
thanthenominalangularvelocityofanobjectatGSO
(∼15″ s
−1
).Underthesecircumstances,wemightexpect
additionaltransientstobevisiblealongthesamenarrowband,
particularlyiftheimagewereextended.Conversely,ifthe
objectspinsslowlyandhasonlyafewsmall,highlyreflective
surfacesdistributedacrossapredominantlynon-reflective
structure,glintsmayoccuronlybrieflyduringtheexposure,
andonlyatspecificorientations.
Toexplorethisfurther,weusetheopen-sourcegraphics
engineBlender
22
tosimulatehowvarious3Dshapescould
produceglintingpatternssimilartothoseobserved.Wemodel
fivedistinctgeometries:asphere,amulti-facetedpolyhedron,
acone,adoublepyramid,andastructurewithtworeflective
panels.Eachshapeiscomposedprimarilyofnon-reflective
material,withlimitedflatsurfacescapableofproducingstrong
specularreflectionswhenorientedpreciselybetweenthe
observerandtheSun.Inadditiontorotation,weallowfor
precessioninsomemodels,whichmodulatesthevisibilityand
timingofglints.Thefivetestgeometriesareshownin
Figure12.
Asexpected,apurelysphericalobjectdoesnotgenerate
short,distinctglints;flat,mirror-likesurfacesarerequired.In
theconemodel,weassumethatthatthetopandbottom
surfacesarereflective,yieldingdoubleglintsperrotation
cycle.Addingprecessionfurtherrestrictsglintvisibility,
producingonlyafewobservableflashesperexposure.
Thedoublepyramidmodelillustratesanotherplausible
case:areflectivestructurethatbecomespartiallydegraded
overtime,leavingonlysmallreflectiveregions.Withrotation
andprecession,suchobjectsmayproduceintermittentglints,
consistentwithwhatweobserveinthedata.
Overall,wefindthateachofthefivetestshapes—under
specificassumptionsregardingspin,precession,andreflective
surfacecoverage—can,inprinciple,reproduceaglinting
patterncompatiblewiththetransientsobservedinPOSS-I
images.
Thegeometricmodelspresentedinthissectionareintended
todemonstratetheplausibilityofproducingalignedglint
patternsfromtumblingorprecessingobjectsinhigh-altitude
orbits.Weemphasizethatthesemodelsareillustrativerather
thanpredictive,andnoattemptismadetofitthespecifictime
separationsorangularoffsetsoftheindividualcandidates.
Whilenoclearperiodicityhasbeenidentifiedinthecurrent
POSS-Idata,itiswellknownfrommodernshort-exposure
surveysthatsomeEarth-orbitingobjects,includingthosein
GSO,canproduceisolated,PSF-likeglintswithoutclear
repetitionpatterns(e.g.,Niretal.2021). Thislackof
periodicitymayresultfromslowrotation,irregularshapes,
orspecificphase-angleconstraintsthatproduceonlyafew
observableflashesperorbitalcycle.
Additionally,theobservedskydistributionofaligned
transientsdoesnotalwaysfollowasimplegreat-circle
geometry,whichcouldreflectthepossibilityofcomplex
trajectories,attitudedrift,oreventhepresenceofmultiple
independentobjects.Poweredobjectscouldevenchangetheir
altitudesortrajectories.Weacknowledgetheseuncertainties
andnotethatmoredetailedmodelingwouldberequiredto
establishstrongerconstraintsonorbitalparametersorglint
periodicity.
9.2.TheBackgroundDensity
Weusethe106,339transientswithdecl.>0°toestimatea
roughdetectionrate.Wealsocountthetransientswithin2°
fromthecentertoavoidpotentialdefectsontheedgeofthe
plate(22,314transients). About635plateswith50minutes
exposuresonaverage,correspondto529hrandanaverage
transientrateof36.3transientsperplatecircleof2°radius.
BasedonthedeficitintheEarthShadowtestinSection8,
roughly1/3rdofthetransientsappeartobeduetosolar
reflections,soroughly12.1transientsperplate.Wecansimply
normalize12.1overtheexposuretimeandplatecoverage
(12.57sq.deg), whichgivesus∼1.1transientshr
−1
deg
−2
of
flux-dilutedflashes,visibledowntor∼20mag.Thistransient
rateisthesumoftransientsonallaltitudesanddistancesfrom
Earth.
Nevertheless,thiscomparisonshouldbeviewedasan
approximate,order-of-magnitudecontrastasthethenumberof
transientsperplateisveryapproximate.Also,modernshort-
exposuresurveyssuchasZTFoperateundervastlydifferent
conditions—usingCCDs,automatedpipelines,andmillise-
cond-leveltimeresolution—whereasthePOSS-Itransients
wererecordedonphotographicplateswithlongintegration
timesandaresubjecttodifferentdetectionbiasesandfalse
positiverates.Arigorouscomparisonwouldrequiremodeling
ofcompleteness,instrumentsensitivity,andeventclassifica-
tioncriteria,whichisbeyondthescopeofthisstudy.
Wecanalsocalculatetheactualnumberdensityofobjects.
Ifweassumethatthepopulationofobjectshasauniform
numberpersurfaceunit(n),thenthenumberofobjects(N)
detectableatanygiventimeisgivenby:()=×NnS, 3
whereSisthesphericalsurveysurfacecontainingtheobserved
reflectiveobjects:()=n
N S
, 4
22
www.blender.org
18
PublicationsoftheAstronomicalSocietyofthePacific,137:104504(22pp), 2025October Villarroeletal.

19. andthesurfaceareaS=2πd
2
iscalculatedforthesunlit
hemisphereattheradiusofaGSOd.Thus:()=n
N d2
. 5
2
Wesetd=42,164kmastheradiusoftheGSO.
Usingtheglintdetectionrateofapproximately∼1.1
transientsperhourpersquaredegree(assumingonedetectable
glintperobjectperhour):()×n 2.010km.
6 all
6 2
Theseestimatesprovideaguidetothesurfacenumber
densityofobjects.However,noteveryobjectwillproduce
severalglints.Sincesomeobjectsmightproducemorethan
oneglint(seeSection3.4inNiretal.2021), wecanassume
thatoneobjectmightproducefrom5to20glints.Theshape
andthereflectivityofanobjectwilldeterminethelikelihood
foroneormoreglints.Thisuncertaintyalsoleadstoan
underestimationofthenumberdensityofobjects,whichcould
actuallybeevenoneorderofmagnitudehigher.Thesurface
densityconstraintquotedhereisafirst-orderestimatebasedon
oureventdetectionrateandassumedskycoverage.These
estimatesdonotincludeafulltreatmentofincompleteness,
observationalbias,orformalstatisticalconfidencelevels,and
shouldthereforebeinterpretedasanindicativeupperbound
ratherthanarigorouslimit.Moreover,thetruefractionoffalse
positivesinthelargersamplefromSolanoetal.(2022), dueto
platedefectsorotherinstrumentalartifacts,remainsunknown.
WhiletheoverallstatisticaltestinSection8isrobusttothis
uncertainty,theabsolutenumberdensityn
all
inferredhere
shouldbeinterpretedwithcautionuntilafullvalidationofthe
samplehasbeenperformed.
10.Discussion
AretheresignaturesofartificialobjectsinEarth’sorbitin
pre-Sputnikimages?Thisisthecentralquestionexploredin
thepresentstudy.Weadoptastraightforwardstrategy:
searchingformultipletransientsalignedalonganarrowband
withinlong-exposurephotographicplatesfromaperiodprior
Figure12.Simulatedshapes.Weshowfivedifferentshapesthatunderslowspinningcouldproduceahandfulofglintsandinparticulardoubleglints.Eachshape
hastwohighlyreflectivesurfaces.Fromtoptobottom:(a)cone-likeshape,(b)multifacesshape,(c)sphere,(d)3Dhexagone,(e)pieceofdebris.Eachobjecthas
bothdullandreflectivematerialsonitssurface,paintedingrayrespectivelighttones.Eachobjectspinsaroundanaxisthatalsohasprecession,causingthereflective
surfacenottobevisibleatalltimes.
19
PublicationsoftheAstronomicalSocietyofthePacific,137:104504(22pp), 2025October Villarroeletal.

20. toknownartificialsatelliteactivity.Thisapproachfollowsthe
principleofseekingNTAsviadistinctive,low-probability
observationalsignatures,or“smokinggunindicators”(Villarroel
etal.2021). UsingthepublishedcatalogofVASCOtransientsin
thenorthernhemisphere(Solanoetal.2022), weidentify∼83
initialcandidater-pointalignments,alongwithalargernumber
ofdoubleandtripletransientgroupings.Tripletsareofparticular
interest,astheyareconsistentwithreflectionsfromflat,rotating
surfaces(Deiletal.2009). Onesuchexamplewaspreviously
reportedinSolanoetal.(2023).
Wemanuallyinspectall22candidatealignmentscontaining
fourormoretransients(notingthatsomereducetothreeafter
closeranalysis), andhighlightthefivemoststatistically
significantcasesinSection6.Althoughtheuncertaintiesdo
notallowustocomputeatotaloccurrenceprobabilityforsuch
alignmentsacrosstheentiresurvey,wedoestimatethechance
probabilityofeacheventwithinasingleimagefield.These
estimates—dependentonassumptionsaboutpoint-spread
functionwidths—yieldsignificancelevelsrangingfrom2.5σ
to4σforthemostpromisingcases.Notably,threecandidates
withfourormorealignedpointsemergeasespeciallystrong,
althoughtwoofthemshowminormorphologicalirregularities.
ThesearenotrelatedtoFWHMdifferences,whichare
addressedseparatelyintheliterature(Villarroeletal.
2025a). Whilewecannotfullyexcludethepossibilityofrare
PSF-likeopticalghostsproducingthealignments,wenotethat
nosimilarfeaturesappearontheblueplatestakenwithsimilar
configurations.
Amongtheremaining3-pointalignments(61intotal), some
mayalsomeritfollow-upifconfirmedasgenuinetransients.
Traditionally,thiswouldrequiremicroscopicexaminationof
theoriginalplates.However,ourdiscoveryofastatistically
significant(>3σ) temporalcorrelationbetweenVASCO
transientsandindependenthistoricalreportsofUAPs(Bruehl
&Villarroel2025) offersadditionalsupportfortheauthenti-
cityofthetransients.Platedefectsorscanningartifactsare
expectedtooccurrandomlyintime;thefactthatthese
transientalignmentsappearpreferentiallywithinadayof
reportedUAPeventsstronglydisfavorsinstrumentalor
spuriousorigins.Inthislight,thecorrelationitselfprovides
indirectbutmeaningfulvalidationofthetransients’reality—
thusreducingthenecessityofmicroscopicinspectionasthe
onlypathtoconfirmation.
Butmostimportantly,Section8presentsacriticaltestofthe
glintinterpretation:wefindastrongdeficitoftransient
detections,atthe∼22σ statisticalsignificancelevel,withinthe
Earth’sumbralshadow.Thisisconsistentwiththeideathat
sunlightisrequiredtoproducetheobservedflashes.Ifthese
eventsaresunlightreflectionsofforbitingobjects,theyshould
vanishintheshadowconeoftheEarth—exactlywhatwe
observe.Thislendssubstantialsupporttotheinterpretation
thatthetransientsarerealastrophysicalornear-Earthevents,
andnotplatedefectsoropticalghosts.Thedisappearanceof
thepopulationinEarth’sshadowwouldnotbeexpectedfor
emulsionflawsorchemicalirregularities.Thesameholdstrue
foropticalghosts.
OfparticularinterestisCandidate5,whichoccurredon
1952July27—thesecondweekendofthewidelydocumented
WashingtonD.C.“UFOflap.”Thiswaveofsightingsinvolved
numerousradardetectionsandpilotobservationsovertwo
consecutiveweekends,July18–19and26–27.Coincidentally,
Candidate1alsooccurredwithinonedayofthepeakofthe
1954UFOwave.ThetripletransientreportedinSolanoetal.
(2023) fallsonthefirstweekendoftheWashingtonevent.
Importantly,thesecandidateswereanalyzedbeforetheauthors
becameawareoftheirproximitytoUAPreports,helpingto
minimizecognitivebias.
Additionally,acorrelationhasbeenfoundbetweenVASCO
transientsandhistoricalnucleartestdates(Bruehl&
Villarroel2025), echoingpaststatisticalstudieslinking
nuclearactivitytoincreasedUAPreports,seee.g.,review
byKnuthetal.(2025). Whilecausalityremainsundetermined,
theconvergenceoftheseindependentcorrelationssuggests
thattheVASCOtransientsarenotrandomartifacts,but
potentiallylinkedtophysicalphenomenaworthyoffurther
investigation.
Usingthetheoreticalframeworkoutlinedearlier,we
simulateglintingpatternsfromplausibleobjectshapesin
GSO.Theseincludemultifacetedandpartiallyreflective
objectswithslowspinsandprecessingaxes.Theinferred
surfacedensityofdetectableobjectsis2.0×10
−6
km
−2
,
thoughthisestimateissubjecttouncertaintybothfrom
unknownshapeandreflectivityfactors(whichmaycause
underestimation), andfromtheunknownfractionoffalse
positivesinthesample(whichmaycauseoverestimation). Itis
worthconsideringwhethersuchapopulationcouldbe
associatedwiththeso-calleduncorrelatedtargets(UCTs),
whichhavebeenconsistentlyreportedinsubstantialnumbers
bymilitaryopticalsensorsandradarssincetheearly1960s.
23
AlthoughtheGSOhypothesisisconsistentwiththedata,no
clearevidenceforperiodicorquasi-periodicglintinghasyet
beenidentified.Objectsspinningslowlyorpossessing
complexreflectivegeometriesmayproduceonlyafewflashes,
complicatingeffortstoestablisharepeatingsignature.More-
over,itremainspossiblethatsomeeventsextendbeyondthe
fieldofviewofasingleplate.Anobjectmovingat∼10″ per
secondcouldtraverseupto10°duringa50minutesexposure,
suggestingthepossibilityoflongeralignmentchainsthan
thosecapturedhere.
Conversely,ifalltransientsweretobeconfirmedasfalse
positives—e.g.,duetorarebutstar-likephotographicplate
artifacts—oursearchstillconstitutesameaningfulupperlimit
23
Forexample,seedeclassifiedU.S.DepartmentofDefensedocuments
releasedunderFOIA,availableviaTheBlackVault:https://documents2.
theblackvault.com/documents/dtic/FOIA2014-128-GC8000301.pdf.
20
PublicationsoftheAstronomicalSocietyofthePacific,137:104504(22pp), 2025October Villarroeletal.

21. onthedensityofNTAsinthenear-Earthenvironment.Inthis
scenario,wederivearoughsurfacedensityconstraintof
<10
−6
objectskm
−2
forhigh-altitudeorbitsintheEarth’s
vicinity(thousandstohundredsofthousandsofkilometers),
evenifthislimitmustbecarefullyapproachedduetothelack
ofmodelingforbiasandincompleteness.Thus,regardlessof
interpretation,ourfindingsprovidenewconstraintsonpossible
technosignaturesnearEarth.
Futureworkshouldconsidersearchingfor“dashed-line”
alignmentsoverlargerplateregions,andinvestigatingsubtle
elongationeffectsinhigh-resolutiondigitizations.Such
elongationscouldindicatemotionacrosstheskyorlarge
objectsize,especiallyifconsistentwiththealignment
direction.
Insummary,wehavepresentedasmallbutcompellingset
ofalignedtransientcandidatesfromapre-satelliteerasky
survey.Whiletheultimateexplanationremainsuncertain,the
convergenceofspatialalignment,statisticalsignificance,and
temporalcorrelationwithindependentaerialanomalyreports
supportstheviewthattheseeventsarelikelyreal—andmay
representaclassofastronomicalphenomenanotyetunder-
stood.AlternativeexplanationsarediscussedinSection7.
11.Conclusions
Thispaperpresentsafirstsystematicsearchformultiple,
simultaneouslyappearingandvanishingopticalpointsources
onlong-exposurephotographicplatesthatalsoexhibitspatial
alignment.WefocusontheredPOSS-Iplates,andpresentfive
topcandidateeventswiththreeormoretransientsaligned
alonganarrowband.Themoststatisticallysignificantcase
(Candidate5)coincidesintimewiththewell-documented
WashingtonD.C.1952UFOflap—oneofthemostprominent
masssightingsofUAPsinrecordedhistory.Aseparatestudy
(Bruehl&Villarroel2025) confirmsastatisticallysignificant
(>3σ) temporalcorrelationbetweenVASCOtransientsand
independenthistoricalUAPreports.
Theoriginofthetransientsremainsunknown.Oneplausible
explanationisthattheyarecausedbybrieflightemissions
fromartificialobjectsinorbitorbyobjectswithanomalous
movementsinEarth’satmosphere—emissionssobriefthat
theyappearaspointsourcesratherthanstreaks,despitethe
telescopetrackingthestars.Alternatively,theycouldarise
fromsolarreflectionsoffflat,highlyreflectivesurfacesat
geosynchronousaltitudes.Thelatterinterpretationisfurther
supportedbyourshadowtestinSection8,whichrevealsa
significantdeficitofsucheventswithintheEarth’sumbra,
consistentwithasolarreflectionoriginanddifficultto
reconcilewithmanyexplanations,includingphotographic
platedefects.
Ourresultsmotivatecontinuedinvestigationofhistorical
skysurveysandtheapplicationofsimilaralignment-based
detectionmethodstomoderndeep-skyimaging.Whetheror
nottheseeventsultimatelypointtotheexistenceofNTAs,the
identificationofstatisticallyimprobable,spatiallyaligned
transientsinpre-satellitedatarepresentsanovelobservational
anomalydeservingoffurtherscientificattention.Futurework
mayhelpclarifywhetherthesetransientsconstituteanew
classofastronomicalphenomena—orrepresentthefirsthints
ofartificialactivitynearourplanet.
Acknowledgments
B.V.wishestothanktheanonymousreferee,thatgave
challengesandsuggestionsthatgreatimprovedthiswork.B.V.
wishestothankDennisÅsbergforhissupportandfor
inspiringdiscussionsaboutthework.Shealsowishestothank
DaveAltmanforteachingherabouttheWashington1952
UFOflap.B.V.alsowishestothankGeoffMarcy,AviLoeb
(Galileoproject), RobertPowell(SCU/Galileo), SarahLittle
(SCU/Galileo) forhelpfulandconstructivecomments.
TheDigitizedSkySurveyswereproducedattheSpace
TelescopeScienceInstituteunderU.S.GovernmentgrantNo.
NAGW-2166.Theimagesofthesesurveysarebasedon
photographicdataobtainedusingtheOschinSchmidtTele-
scopeonPalomarMountainandtheUKSchmidtTelescope.
Theplateswereprocessedintothepresentcompresseddigital
formwiththepermissionoftheseinstitutions.TheNational
GeographicSociety—PalomarObservatorySkyAtlas(POSS-
I)wasmadebytheCaliforniaInstituteofTechnologywith
grantsfromtheNationalGeographicSociety.TheSecond
PalomarObservatorySkySurvey(POSS-II) wasmadebythe
CaliforniaInstituteofTechnologywithfundsfromthe
NationalScienceFoundation,theNationalGeographic
Society,theSloanFoundation,theSamuelOschinFoundation,
andtheEastmanKodakCorporation.TheOschinSchmidt
TelescopeisoperatedbytheCaliforniaInstituteofTechnology
andPalomarObservatory.TheUKSchmidtTelescopewas
operatedbytheRoyalObservatoryEdinburgh,withfunding
fromtheUKScienceandEngineeringResearchCouncil(later
theUKParticlePhysicsandAstronomyResearchCouncil),
until1988June,andthereafterbytheAnglo-Australian
Observatory.TheblueplatesofthesouthernSkyAtlasand
itsEquatorialExtension(togetherknownastheSERC-J), as
wellastheEquatorialRed(ER), andtheSecondEpoch[red]
Survey(SES) werealltakenwiththeUKSchmidt.Alldataare
subjecttothecopyrightgiveninthecopyrightsummary.
Copyrightinformationspecifictoindividualplatesisprovided
inthedownloadedFITSheaders.Supplementalfundingfor
sky-surveyworkattheSTScIisprovidedbytheEuropean
SouthernObservatory.
ThisresearchhasmadeuseoftheSpanishVirtual
Observatory(http://svo.cab.inta-csic.es) supportedfromMin-
isteriodeCienciaeInnovaciónthroughgrantPID2020-
112949GB-I00.B.V.isfundedbytheSwedishResearch
Council(Vetenskapsrådet,grantNo.2024-04708)andisalso
21
PublicationsoftheAstronomicalSocietyofthePacific,137:104504(22pp), 2025October Villarroeletal.

22. supportedbyananonymousdonortowhomsheisdeeply
grateful.M.E.S.acknowledgesfinancialsupportfromthe
AnnieJumpCannonFellowship,supportedbytheUniversity
ofDelawareandendowedbytheMountCubaAstronomical
Observatory.
DataAvailability
Datawillbesharedonreasonablerequesttothecorresp-
ondingauthor.
ORCIDiDs
AlinaStreblyanska aahttps://orcid.org/0000-0001-8876-9102
References
Andruk,V.,Eglitis,I.,Protsyuk,Y.,etal.2019,OAP, 32,181
Andruk,V.,Kharchenko,N.,Schilbach,E.,&Scholz,R.-D.1995,AN,
316,225
Andruk,V.M.,Pakuliak,L.K.,Golovnia,V.V.,etal.2017,SciIn, 13,17
Bracewell,R.N.1960,Natur, 186,670
Bruehl,S.,&Villarroel,B.2025,ScientificReports,inpress,https://www.
researchsquare.com/article/rs-6347224/v1
Corbett,H.,Law,N.M.,Soto,A.V.,etal.2020,ApJL, 903,L27
Deil,C.,Domainko,W.,Hermann,G.,etal.2009,APh,31,156
Edmunds,M.G.1981,Natur, 290,481
Edmunds,M.G.,&George,G.H.1985,MNRAS, 213,905
Freitas,R.A.,Jr.,&Valdes,F.1980,Icar,42,442
Freitas,R.A.,Jr.,&Valdes,F.1985,AcAau, 12,1027
Greiner,J.,Wenzel,W.,&Degel,J.1990,A&A,234,251
Grindlay,J.,Tang,S.,Los,E.,&Servillat,M.2012,inIAUSymp.285,New
HorizonsinTime-DomainAstronomy,Proc.Int.AstronomicalUnion
(Cambridge:CambridgeUniv.Press), 29
GuyNir’scode2024,https://github.com/guynir42/earthshadow
Hambly,N.C.,&Blair,A.2024,RASTI, 3,73
Hambly,N.C.,MacGillivray,H.T.,Read,M.A.,etal.2001,MNRAS,
326,1279
Haqq-Misra,J.,&Kopparapu,R.2012,AcAau, 72,15
Irureta-Goyena,B.Y.,etal.2025,PASP, 137,054503
Knuth,K.H.,Ailleris,P.,Agrama,H.A.,etal.2025,ProgressinAerospace
Sciences, 156,101097,https://www.sciencedirect.com/science/article/
abs/pii/S0376042125000235
Nir,G.,Ofek,E.O.,Ben-Ami,S.,etal.2021,MNRAS, 505,2477
Solano,E.,Villarroel,B.,Rodrigo,C.,etal.2022,MNRAS, 515,1380
Solano,E.,Marcy,G.,Villarroel,B.,etal.2023,MNRAS, 527,6312
Teodorani,M.2004,JournalofScientificExploration,18,217
Tokovinin,A.2002,PASP, 114,1156
Valdes,R.A.,&Freitas,F.,Jr.1983,Icar,53,453
Vavilova,I.B.,Pakulyak,L.K.,Shlyapnikov,A.A.,etal.2012,KPCB,
28,85
Vavilova,I.B.,Yatskiv,Y.S.,&Pakuliak,L.K.2017,inIAUSymp.325,
Astroinformatics,Proc.Int.AstronomicalUnion(Cambridge:Cambridge
Univ.Press), 361
Villarroel,B.2024,TheVanishingStarEnigmaandthe1952WashingtonD.
C.UFOWave(PopularScience)
Villarroel,B.,Imaz,I.,&Bergstedt,J.2016,AJ,152,76
Villarroel,B.,Marcy,G.W.,Geier,S.,etal.2021,NatSR, 11,12794
Villarroel,B.,Mattsson,L.,Guergouri,H.,etal.2022a,AcAau, 194,106
Villarroel,B.,Pelckmans,K.,Solano,E.,etal.2022b,Univ., 8,561
Villarroel,B.,Solano,E.,&Marcy,G.W.2025a,arXiv:2507.15896
Villarroel,B.,Soodla,J.,Comerón,S.,etal.2020,AJ,159,8
Villarroel,B.,Watters,W.A.,Streblyanska,A.,Solano,E.,&Geier,S.2025b,
MNRAS,inpress
Waszczak,A.,Prince,T.A.,Laher,R.,etal.2017,PASP, 129,034402
22
PublicationsoftheAstronomicalSocietyofthePacific,137:104504(22pp), 2025October Villarroeletal.