17th District OSI Cumulative Sighting Log of Unknown Aerial Phenomena, December 1948 – May 1950 (Full Read, Sightings 1–204)

Summary

This is the full read of the 17th District OSI Cumulative Sighting Log of Unknown Aerial Phenomena that pass 11 (belmont-twinkle-master-memo-osi-log-1949-1950) sampled through four slice pages. The log is the structured-table backbone behind A. H. Belmont’s August 23, 1950 master memo that recorded “approximately 150 observations of aerial phenomena” in the vicinity of New Mexico installations since December 1948. The actual numbered entries run higher than the Belmont count: the OCR’d portions of Section 6 confirm at least 204 numbered sightings (the highest entry read is #204, Kirtland AFB, March 22, 1950, on page 70). The log is dated 25 May 1950 and is signed out by 17th District OSI under Lt. Col. Doyle Rees with distribution to USAF Director of Special Investigations, AMC Wright-Patterson, Special Weapons Command, AEC Los Alamos, and FBI Albuquerque/El Paso (page 26).

The full read covers a date range of 18 January 1948 (entry #1, USSR Brest-Kiev impact line) through 22 March 1950 (entry #204, Kirtland AFB) — twenty-six months of structured AF/OSI UAP observation. The cumulative-log format is unique in the 62-HQ-83894 archive: eighteen-column structured table per sighting (Number, Date, Time, Number of Observers, Reliability, General Area, Apparent Direction of Flight, Apparent Altitude, Course, Color, Train or Trail, Duration, Sound, Shape, Apparent Size, Apparent Speed, Manner of Disappearance, Evaluation). It is the most operationally formatted UAP-data record in the entire archive read so far — closer to a modern incident-reporting database than to a narrative case file.

Geographic concentration is overwhelming and directly reflects the AAR/Vital-Installations geography documented in pass 9 (project-grudge-vital-installations-1948-1949). Of the entries read, the dominant cluster is Los Alamos (well over 30 entries), with Sandia Base, Albuquerque, Alamogordo, Killeen Base, Camp Hood, Holloman AFB, Roswell, Tucson, White Sands, and Vaughn making up nearly all of the remainder. The log is essentially a perimeter-watch record of the New Mexico nuclear-installation corridor, plus the Camp Hood / Killeen Base nuclear-weapons-storage installation in Texas, plus a thin Arizona ring (Tucson, Davis-Monthan, Nogales, Douglas, Winslow, Flagstaff, Phoenix). One entry (#1) is offshore — the Brest-Kiev line of impact in the USSR, January 1948 — which the log includes as the calibration baseline before the New Mexico data picks up in late 1948.

The temporal distribution shows two clusters and a sustained tail. December 1948 through January 1949 is the original Green Fireball wave centered on Los Alamos (entries 12–18). March through May 1949 is a Camp Hood / Killeen Base / Los Alamos run with multiple multi-day-watching observations (entries 35–67). August through October 1949 is the heaviest operational stretch with sustained Killeen Base traffic and a major October 6, 1949 multi-witness multi-site cluster (entries 124–128 — five sightings of a green object across Alamogordo, Albuquerque, Mescalero, and Wagon Mound, all 1750–1800 hrs). February through March 1950 is the Project Twinkle setup period with the Los Alamos February 25 cluster (entries 179–187) and the late-March Sandia/Kirtland surge (entries 200–204). The phenomena classified as “discs” gradually replace green-fireball-classified entries as the log moves into 1950 — by the spring of 1950 the predominant evaluation code is (2) Disk or Variation, not (1) Green Fireball.

The most witness-dense single date in the full read is still February 25, 1950 at Los Alamos: entries 180, 184, 185, 186, 187 — five separate sighting numbers between 1545 and 1655 hrs, including #186’s fourteen reliable observers (the apex case in the entire log). The October 11, 1949 Roswell entry (#134) records six observers tracking a green-to-orange object that “maneuvered up & down” for 45 minutes. The February 18, 1950 Holloman AFB entry (#171) records five reliable observers tracking a white-to-orange round-to-cone-shape object for 1 hour 44 minutes — the longest duration in the entire log read, longer than the Vaughn 1 hr 25 min entry (#189) noted in pass 11. The El Paso January 30, 1949 entry (#18) records “App 200” observers tracking a green ball that “broke into pieces.”

What the OSI Log Documents (Aggregate View)

Time distribution

The log’s entries cluster in three identifiable runs:

December 1948 – February 1949 (entries 5–34, mostly missing from OCR but bracketed by entries 12–18 on page 36): the original Green Fireball wave, dominated by Los Alamos, Las Vegas (NM), and El Paso. Entry #18 (January 30, 1949, El Paso, “App 200” observers) is the largest-witness-count entry in the entire log read.
March – May 1949 (entries 35–67, pages 40–45): a Camp Hood / Killeen Base run interleaved with Los Alamos. Multi-night observation of the same diamond-shape object at Camp Hood across May 7–8 (entries 64, 65, 66) is documented as a multi-night-tracking pattern.
August – November 1949 (entries 90–152, pages 50–60): the heaviest sustained run. Killeen Base / Camp Hood traffic in August (entries 94, 95, 100–105). Multi-witness multi-site Sandia Base cluster on September 27, 1949 (entries 115, 116, 117, 118, 119 — five separate entries between 0130 and 0300 hrs). Cross-state cluster October 6, 1949 (entries 122–128 — seven separate entries, 1745–1800 hrs, across Mescalero, Alamogordo, Albuquerque, Wagon Mound — all coordinated within fifteen minutes).
December 1949 – March 1950 (entries 153–204, pages 61–70): the Project Twinkle setup period. Holloman AFB enters the log heavily (entries 171, 172, 190). The Los Alamos February 25, 1950 cluster (entries 179, 180, 184–187) is the apex multi-witness day. The Kirtland AFB March 21–22 surge (entries 200–204) is the immediate pre-Twinkle period.

Installation clustering

The log entries occur primarily AT Vital Installations, not merely NEAR them. This is operationally significant: the OSI was logging witness reports from people stationed at the installations, not from civilian observers in the surrounding desert. Specifically:

Los Alamos, New Mexico (atomic-weapons design lab): entries 14, 15, 17, 36, 37, 47, 48, 49, 50, 59, 60, 61, 63, 71, 72, 73, 74, 110, 112, 121, 130, 139, 141, 143, 144, 145, 146, 147, 148, 155, 179, 180, 183, 184, 185, 186, 187, 193, 194 — well over 30 entries. Los Alamos accounts for roughly a quarter of the entire log.
Sandia Base, New Mexico (atomic-weapons storage): entries 113, 115, 116, 117, 118, 119, 120, 131, 132, 133, 170, 200, 201, 202.
Killeen Base, Texas (atomic-weapons storage at Camp Hood): entries 94, 95, 105 — independent from the surrounding Camp Hood entries (which run 38, 39, 46, 62, 64, 65, 66, 100, 102, 103). Together Killeen Base + Camp Hood is the second-largest installation cluster after Los Alamos.
Holloman AFB / Alamogordo / White Sands: entries 55 (White Sands, 5 observers VR), 69, 92, 93, 106, 124, 125, 135, 137, 153, 171, 172, 190.
Kirtland AFB: entries 169, 203, 204.
Davis-Monthan AFB: entry 109.

The pattern: every major Cold War nuclear-installation site in the Southwest is in the log. The geography is not random — it is the post-Manhattan-Project nuclear-weapons-production-and-storage perimeter.

Multi-witness entries

Entries with 4 or more observers in the full read:

#18, January 30, 1949, El Paso, “App 200” Unk reliability — green ball, broke into pieces. The largest-witness-count entry in the log, but reliability rated Unknown.
#186, February 25, 1950, Los Alamos, 14 observers Reliable — silver round, fluttering motion, 500–1500 mph. The fourteen-observer apex documented in pass 11.
#191, March 16, 1950, Farmington, NM, 10 observers Unk — bright luminous tin foil flat spheric, 3–5 mts.
#192, March 16, 1950, Farmington, NM, 10 observers Unk — bright aluminum oval and oblong, 30 mts. Two ten-observer entries from the same Farmington daytime sighting on the same day (the “Farmington Armada” of March 17, 1950 — though here logged as the 16th).
#180, February 25, 1950, Los Alamos, 12 observers Reliable — flashing silver circular like plane fuselage, 3 secs to 2 mts. A separate twelve-observer entry from the Los Alamos February 25 cluster, distinct from #186.
#204, March 22, 1950, Kirtland AFB, 11 observers Unk — flying-wing-shape, tan to brown, “extremely high speed.”
#188, March 10, 1950, Phoenix, 6 observers Unk — aluminum or quick-silver oval, moved upward at 60° angle.
#134, October 11, 1949, Roswell, 6 observers Unk — green-to-orange round, “maneuvered up & down” for 45 mts.
#13, December 12, 1948, 15 mi south of Las Vegas NM, 5 observers Very Reliable — bright green ball, broke into 3 or 4 small fragments.
#55, April 24, 1949, White Sands, 5 observers Very Reliable — white-light yellow ellipsoid, “tremendous rate of speed,” 60 secs.
#108, August 21, 1949, Nogales AZ, 5 observers Unk — dull orange wafer, “10 times speed of jet planes.”
#148, November 25, 1949, Los Alamos, 5 observers Reliable — yellowish green, downward 10° from horizontal.
#171, February 18, 1950, Holloman AFB, 5 observers Reliable — white and orange round to cone shape, 1 hr 44 mts duration.
#189, March 5, 1950, Vaughn, NM, 4 observers Reliable — white round, 1 hr 25 mts duration (documented in pass 11).
#14, December 20, 1948, Los Alamos, 4 observers Reliable — pale green / bluish white ball, basketball size.
#64, May 7, 1949, Camp Hood, 4 observers Reliable — green-white diamond shape, 40 mts.
#67, May 8, 1949, Tucson, 4 observers Unk — metallic circular, 40–75 ft diameter, climbed at 45° angle until out of sight.
#138, October 12, 1949, Tucson, 4 observers Reliable — white or silver round, 30,000’, 1,000 mph.
#203, March 21, 1950, Kirtland AFB, 4 observers Unk — bright silver round, zig-zag motion up and down.

Sixteen distinct entries with 4 or more witnesses. Multi-witness simultaneous reports are a baseline pattern in the log, not exceptional.

Long-duration entries

Entries with duration over 5 minutes (durations are the most evidentially-substantive observations because they exceed the natural duration of a meteor by orders of magnitude):

#171, February 18, 1950, Holloman AFB, 1 hr 44 mts — five reliable observers (longest in the log read).
#189, March 5, 1950, Vaughn NM, 1 hr 25 mts — four reliable observers.
#62, May 6, 1949, Camp Hood, 50 mts — alternating pinkish-to-green round, 1200’ dropping to 440’.
#134, October 11, 1949, Roswell, 45 mts — six observers, light-green-turned-orange, “maneuvered up & down.”
#64, May 7, 1949, Camp Hood, 40 mts — four reliable observers, diamond shape, 1300’.
#142, October 21, 1949, Roswell, 20–30 mts — very bright white, 30° above horizon.
#42, March 27, 1949, Tucumcari, 25 mts — amber long-and-narrow, 30° above horizon.
#172, February 20, 1950, Holloman AFB, 5 mts — two observers, white, stationary at 1000’ above 9000’ mountain.
#192, March 16, 1950, Farmington, 30 mts — ten observers, bright aluminum oval and oblong.
#201, March 21, 1950, Sandia Base, 30 mts — silver-to-shady-gray round, 40,000–60,000 ft.
#52, April 15, 1949, El Paso, 15 mts — grayish, “very thin smoke trail,” “Dissipated.”
#44, March 27, 1949, Tucumcari, 15 mts — bright orange long-and-narrow.
#45, March 27, 1949, Tucumcari, 15 mts — orange-fire kite-tail shape.
#190, March 11, 1950, Holloman AFB, 5 mts — color-changing object (light orange to blood red to amber to light green), “ping-pong ball at arm’s length.”
#107, August 20, 1949, Douglas AZ, 6–10 secs notable for 3500–4500 mph speed estimate.
#170, February 15, 1950, Sandia Base, 30 secs — but logged with the unusual color combination “appeared red and green,” round, marble-sized.
#188, March 10, 1950, Phoenix, 10 mts — six observers, aluminum/quick-silver, 40,000–50,000 ft.
#191, March 16, 1950, Farmington, 3–5 mts — ten observers, bright luminous tin foil.

The Holloman AFB 1-hour-44-minute, the Vaughn 1-hour-25-minute, the Camp Hood 50-minute, the Roswell 45-minute, and the Camp Hood 40-minute entries all involve multiple witnesses at military installations. These are the entries hardest to debunk as ordinary meteoric phenomena.

Evaluation-code distribution

The log header defines three Evaluation codes: (1) Green Fireball Phenomena, (2) Disk or Variation, (3) Probable Meteor. No (4)+ codes appear in the OCR’d portions read.

Rough distribution across the OCR’d entries with codes assigned:

Code (1) Green Fireball: dominant in the December 1948 – November 1949 window. Entries 13, 16 (implicit), 35, 49, 51, 58, 59, 60, 61, 62, 63, 64, 65, 66, 71, 72, 73, 74, 91, 92, 93, 108, 121, 122, 124, 125, 126, 127, 128, 129 (implicit), 132, 133, 134, 140, 141, 142 (no, this is code 2), 143, 144, 145, 146, 147, 148, 149, 150, 151, 152, 153, 154, 155, 156, 157, 170 — roughly 40+ entries.
Code (2) Disk or Variation: dominant in the December 1949 – March 1950 window. Entries 41, 42, 43, 44, 45, 50, 52, 54, 55, 56, 57, 67, 68, 69, 70, 107, 109, 114, 135, 136, 137, 138, 139 (no, this is 1), 158, 169 (implicit), 171, 172, 173, 180, 181, 182, 183, 184, 185, 186, 187, 188, 189, 190, 191, 192, 200, 201, 202, 203, 204 — roughly 35+ entries.
Code (3) Probable Meteor: scattered throughout. Entries 38, 39 (implicit), 102, 103, 104, 105, 106, 110 (implicit), 112, 130, 131, 132, 133, 140, 154, 155, 156, 157, 193, 194 — roughly 20+ entries. Many entries carry combined codes (1)/(3) or (2)/(1) — the 17th District analysts allowed an entry to carry two evaluation codes when uncertain.

The taxonomy shift from (1)-dominated in 1948–49 to (2)-dominated in 1950 is the operational signal that pass 9 anticipated: the New Mexico phenomena diversified from green-fireball-only into a broader disc category as the observation window widened.

Notable single entries (not in pass 11)

Sighting #18 (January 30, 1949, El Paso, Texas, App 200 observers Unk, 1754 hrs) — green ball, “Broke into pieces.” The largest single-witness-count entry in the log. (Page 36.)
Sighting #67 (May 8, 1949, Tucson, Arizona, 4 observers Unk, 0930–1100 hrs) — “Metallic circular,” 40–75 feet in diameter, started at 4000 ft and climbed to 20,000 ft. “Motionless to faster than jet.” “Climbed at 45° angle until out of sight.” Daytime, four observers, observed maneuvering from stationary to faster-than-jet vertical climb. Evaluation (2). This is one of the cleanest controlled-flight entries in the log. (Page 45.)
Sighting #107 (August 20, 1949, Douglas, Arizona, 2 observers Unk, 2130 hrs) — “Round to long like inverted saucer,” 8000–10,000 ft, “Flat trajectory,” 6–10 secs duration, 3500–4500 mph speed. Evaluation (2). The fastest speed estimate in the log read. (Page 53.)
Sighting #109 (August 26, 1949, Davis-Monthan AFB, Arizona, 1 observer Reliable, 1345 hrs) — “Similar to triangle with round edges,” 50,000 ft altitude, 60 secs duration, “Terrific rate of speed.” Daytime, single reliable observer. Evaluation (2). The triangle-with-round-edges shape is unusual in the log. (Page 53.)
Sighting #134 (October 11, 1949, Roswell, New Mexico, 6 observers Unk, 2010 hrs) — “Light green turned orange,” round, baseball-sized, “Maneuvered up & down” for 45 mts. Six-witness multi-modal-color-change extended observation at the most-named UAP site in mid-century US history. Evaluation (1). (Page 57.)
Sighting #171 (February 18, 1950, Holloman AFB, 5 observers Reliable, 0510 hrs) — “White and orange,” round to cone shape, “Climbed,” 1 hr 44 mts duration. The longest-duration entry in the log read, observed by five reliable witnesses at an Air Force base. Evaluation (2). (Page 64.)
Sighting #172 (February 20, 1950, Holloman AFB, 2 observers Unk, 0530 hrs) — “Stationary,” white, round, half-dollar at arm’s length, 5 mts at 1000’ above a 9000’ mountain. A stationary UAP at Holloman, two days after entry #171. Evaluation (2). (Page 64.)
Sighting #190 (March 11, 1950, Holloman AFB, 1 observer Reliable, 0100 hrs) — color-changing through “light orange to blood red to amber to light green,” ping-pong ball at arm’s length, 5 mts. The four-stage color shift is the most explicit chromatic-anomaly entry in the log. Evaluation (2). (Page 68.)
Sighting #191 + #192 (March 16, 1950, Farmington, New Mexico, 10 + 10 observers Unk, 1100 + 1000 hrs) — bright tin-foil flat spheric and bright aluminum oval-and-oblong objects, “Faster than conventional aircraft,” 3–5 minutes and 30 minutes durations. Two ten-observer daytime entries on the same day at the same site. The “Farmington Armada.” Evaluation (2). (Page 68.)
Sighting #204 (March 22, 1950, Kirtland AFB, 11 observers Unk, 1100 hrs) — “Flying wing” shape, tan to brown, 25,000–30,000 ft, “Extremely high speed,” 5–6 secs. Eleven witnesses at Kirtland watching what they recorded as a flying-wing-shaped object at high altitude. Evaluation (2). (Page 70.)

Coverage gaps and methodological notes

OCR coverage is incomplete. Within Section 6 the following pages are missing from the converted directory: 35, 37, 38, 39, 47, 48, 49, 51, 62, 63, 65, 69, 71. Based on the entry-numbering pattern (page 36 has #12–18, page 40 has #35–38; page 46 has #68–74, page 50 has #90–95; page 61 has #153–158, page 64 has #169–173; page 68 has #190–194, page 70 has #200–204) — the missing pages account for roughly 50–60 unread entries between #5–11, #19–34, #75–89, #96–104 (page 51 and partial 52), #159–168, #174–178, and #195–199. The full-archive entry count is therefore at minimum 204 (highest read) and likely higher (page 71+ would continue the series).

Reliability codes used: R (Reliable), VR (Very Reliable), Unk (Unknown Reliability). Methodological criteria for assigning the codes are not in the table headers. Visual-pattern distribution: the highest-witness-count entries are about evenly split between R and Unk — the OSI did not appear to discount unreliable-rated entries when those entries had multiple corroborating witnesses.

Some entries are blank-numbered (the OCR shows several rows where the Number column is empty — e.g., page 43 entry 53 between 52 and 54, page 45 entry 64 between 63 and 65). These appear to be transcription artifacts in the OCR rather than actual gaps in the original log.

Some entries have blank Evaluation codes (e.g., page 36 entry 14, 15, 17 — three of the original Green Fireball wave entries are unevaluated despite being among the most-discussed cases in La Paz’s Seventh Report). The evaluation column was filled retroactively, not at intake — implying that as of May 25, 1950, OSI had not finalized classification of every entry in its own log.

Why This Matters

The full read confirms the log runs to at least #204, not #189. The Belmont memo’s “150 observations” undercount is now in primary form: by March 22, 1950, OSI’s own internal log carried 204+ numbered sightings. The Belmont count was either selective (excluding entries Belmont deemed Probable Meteor) or out-of-date (counting only through summer 1950 minus the spring 1950 surge). Either way, the OSI log is broader than the FBI’s own internal summary suggests.
The OSI Cumulative Sighting Log is the most operationally formatted UAP-data record in the entire 62-HQ-83894 archive. Eighteen-column structured table per sighting, sustained across 26 months (Jan 1948 – Mar 1950), distributed to AMC Wright-Patterson, AEC Los Alamos, Kirtland Special Weapons Command, Sandia, Holloman, FBI Albuquerque/El Paso. This is mid-century AF/OSI structured-data UAP intake at its operational peak.
The geographic distribution is the New Mexico nuclear-installation perimeter, not random. Los Alamos accounts for roughly a quarter of the entire log; combined with Sandia Base, Killeen Base, Camp Hood, Holloman AFB, Kirtland AFB, White Sands, and Alamogordo, the AEC/AMC/SWP installation footprint accounts for the overwhelming majority of entries. The pattern is what would be expected if the phenomena were either US classified tests, foreign reconnaissance of those installations, or a non-mundane phenomenon attracted to the same infrastructure.
Multi-witness simultaneous reports are a baseline pattern, not exceptional. Sixteen entries have 4+ observers, including the El Paso ~200-observer entry (#18), the Los Alamos 14-observer cluster (#186), and the two Farmington 10-observer entries (#191, #192). Multi-witness UAP cases are not rare in this dataset — they are routine.
Long-duration entries are not rare either. Five entries exceed 30 minutes, two exceed 1 hour 25 minutes. The Holloman AFB 1 hr 44 mt entry (#171) and the Vaughn 1 hr 25 mt entry (#189) are unambiguous long-duration UAP observations at Air Force installations by reliable multi-witness groups. These are the entries that resist meteor-debunking.
The (1)/(2)/(3) Evaluation taxonomy is the precursor to ATIC’s later UFO classification. The OSI’s Green Fireball / Disc / Probable Meteor categories are the operational taxonomy that fed into Project Sign / Grudge / Blue Book. Combined-code entries (1)/(3) and (1)/(2) document a working classification system in real-time use, not retroactive.
The reliability-rating practice (R / VR / Unk) is mid-century structured-source-evaluation methodology in primary form. The OSI log shows trained observers being categorized by rating and the rating being applied to evaluate their reports — predating the formalization of source-evaluation methodology in later intelligence doctrine.
The October 6, 1949 multi-site cluster (entries 124–128, 1745–1800 hrs) is one of the most coordinated cross-site simultaneous-observation patterns in the log. Five separate sightings in fifteen minutes across four New Mexico towns (Mescalero, Alamogordo, Albuquerque, Wagon Mound) tracking a green-to-greenish-white object. This is geographic coverage that suggests a single transiting object visible across hundreds of miles of New Mexico, not an incoherent witness panic.
The taxonomy shift from (1)-dominated in 1948–49 to (2)-dominated in 1950 is the documentary signal of the New Mexico phenomena evolving from “green fireballs” into “discs” — the same shift Belmont’s August 23, 1950 memo records taxonomically and that Project Twinkle was set up to investigate empirically.
The log is dated 25 May 1950, two days after La Paz’s May 23 Seventh Report. OSI compiled this log immediately following La Paz’s submission, distributing it to USAF / AMC / AEC / FBI within a 48-hour window. The temporal coupling between La Paz’s scientific report and OSI’s structured-data dump is documented.

Connections

PURSUE full inventory
PURSUE master report (master synthesis)
belmont-twinkle-master-memo-osi-log-1949-1950 — pass 11 source page; this page is the full-read companion
la-paz-seventh-report-cabell-directive-twinkle — pass 10; the OSI log is dated two days after La Paz’s May 23 Seventh Report
project-grudge-vital-installations-1948-1949 — pass 9; the OSI log is the structured-data record of sightings AT the Vital Installations the Grudge framework defined
stuart-adcock-oak-ridge-march-1950 — pass 13; another Vital Facilities case
rhodes-phoenix-photographs-1947
UAP disclosure (concept)

Open Questions

The remaining missing OSI log pages (Section 6 pages 35, 37, 38, 39, 47, 48, 49, 51, 62, 63, 65, 69, 71). Approximately 50–60 unread numbered entries between #5–11, #19–34, #75–89, #96–104, #159–168, #174–178, #195–199. The original PDF on war.gov would contain these pages — the OCR’d archive is incomplete.
The full upper bound of the log. The highest entry read is #204 (March 22, 1950). Whether the log continues past #204 in Section 6 pages beyond 70 (or in Section 7) is unverified. The log is dated 25 May 1950 but the final entry on the last continuation page should give the actual cutoff date.
The Sighting #175 photograph (Datil, Cpl Lertis E. Stanfield, February 24–25, 1950, page 33 of Section 6). Page 33 is the photograph context with La Paz’s analysis (“not the moon, not Venus, not a fixed star out of focus”); the corresponding tabular log entry #175 sits in the missing-OCR span between pages 64 and 66. The Stanfield photograph is one of the very few primary-source photographic entries in the log and warrants its own pass.
The Farmington March 16, 1950 cluster (entries #191, #192, two ten-observer entries, both daytime). This is the “Farmington Armada” event, widely discussed in civilian UFO literature. The OSI log dates it March 16, 1950; civilian accounts typically cite March 17. Whether the OSI log date or the civilian-account date is correct is a small but worth-checking discrepancy.
The methodological criteria for the R / VR / Unk reliability ratings. Not documented in the table headers. Whether based on observer training, prior-report consistency, military-rank, or another criterion.
The “Evaluation” assignment workflow. Some entries have blank Evaluation columns despite being from the most-discussed Green Fireball cases (e.g., entries 14, 15, 17). Whether evaluations were assigned at intake or retroactively, and by whom.

Quotes Worth Keeping

“Sighting #18, 30 Jan 1949, 1754 hrs, App 200 observers Unk reliability, El Paso, Texas, NW to SE, 3°–5° above horizon, Horizontal, Green, [no train], [no duration], None [sound], Ball, [no size], [no speed], Broke into pieces.” — 17th District OSI Cumulative Sighting Log (Section 6 page 36, Sighting #18). The largest-witness-count single entry in the log read — approximately 200 observers tracking a green ball that broke into pieces over El Paso. The reliability rating is Unknown but the witness count is documentary.

“Sighting #67, 8 May 1949, 0930–1100 hrs, 4 observers Unk, Tucson, Arizona, W with 90° turn to the N, 4000 to 20,000’, Horizontal then rapid climb at 45° angle, White, [no trail], 10–20 mts, None [sound], Metallic circular, 40–75’ in diameter, Motionless to faster than jet, Climbed at 45° angle until out of sight, (2).” — 17th District OSI Cumulative Sighting Log (Section 6 page 45, Sighting #67). A daytime four-observer report of a 40–75-foot metallic circular object that maneuvered from stationary to faster-than-jet vertical climb over Tucson. One of the cleanest controlled-flight entries in the log.

“Sighting #107, 20 Aug 1949, 2130 hrs, 2 observers Unk, Douglas, Arizona, N to S, 8,000’–10,000’, Flat trajectory, [no color], None [trail], 6–10 secs, None [sound], Round to long like inverted saucer, About size of single engine airplane, 3500 to 4500 mph, Disappeared in distance, (2).” — 17th District OSI Cumulative Sighting Log (Section 6 page 53, Sighting #107). A 3500–4500 mph speed estimate at low altitude — the fastest speed in the log read.

“Sighting #134, 11 Oct 1949, 2010 hrs, 6 observers Unk, Roswell, New Mexico, Appeared moving to N & angling slightly to E, [no altitude], Maneuvered up & down, Light green turned orange, Yes [trail], 45 mts, None [sound], Round, Size of baseball, [no speed], (1).” — 17th District OSI Cumulative Sighting Log (Section 6 page 57, Sighting #134). A six-observer 45-minute extended observation at Roswell of a green-to-orange round object that “maneuvered up and down.”

“Sighting #171, 18 Feb 1950, 0510 hrs, 5 observers Reliable, Holloman AFB, New Mexico, [no direction], [no altitude], Climbed, White and orange, None [trail], 1 hr 44 mts, None [sound], Round to cone shape, Size of coffee cup at arm’s length, [no speed], Stopped observation, (2).” — 17th District OSI Cumulative Sighting Log (Section 6 page 64, Sighting #171). The longest-duration entry in the log read — five reliable observers at Holloman Air Force Base tracking a white-and-orange round-to-cone-shape object for 1 hour 44 minutes.

“Sighting #190, 11 Mar 1950, 0100 hrs, 1 observer Reliable, Holloman AFB, New Mexico, App 270°, About 30° above horizon at distance of 50 miles, Straight flight, Changed from light orange to blood red to amber to light green, None [trail], 5 mts, None [sound], Ping pong ball, Ping pong ball held at arm’s length, [no speed], Disappeared from view, (2).” — 17th District OSI Cumulative Sighting Log (Section 6 page 68, Sighting #190). Four-stage color shift (light orange → blood red → amber → light green) at Holloman over five minutes. The most explicit chromatic-anomaly entry in the log.

“Sighting #204, 22 Mar 1950, 1100 hrs, 11 observers Unk, Kirtland AFB, New Mexico, NW changing to N, 25,000 to 30,000’, Horizontal, Tan to brown, None [trail], 5–6 secs, None [sound], Flying wing, About size of golf ball held at arm’s length, Extremely high speed, Disappeared, (2).” — 17th District OSI Cumulative Sighting Log (Section 6 page 70, Sighting #204). Eleven observers at Kirtland AFB watching a flying-wing-shaped tan-to-brown object at 25,000–30,000 ft.

“DISTRIBUTION: 6 cys, Director of Special Investigations, Headquarters USAF; 1 cy, CG, Air Materiel Command, Wright-Patterson AFB, Ohio, ATTN: Director of Technical Intelligence; 1 cy, CG, Special Weapons Command, Kirtland AFB, New Mexico; 1 cy, CG, Armed Services Special Weapons Project, Sandia Base, New Mexico, ATTN: J-2; 1 cy, CO, Holloman AFB, New Mexico; 1 cy, Director, Security Division, U. S. Atomic Energy Commission, Los Alamos, New Mexico, ATTN: Mr. B. O. Wells; 1 cy, Federal Bureau of Investigation, El Paso, Texas; 1 cy, Federal Bureau of Investigation, Albuquerque, New Mexico.” — Section 6 page 26, distribution block of the May 25, 1950 OSI Summary. The OSI Cumulative Sighting Log was distributed simultaneously to USAF Director of Special Investigations, AMC Wright-Patterson Director of Technical Intelligence, Special Weapons Command Kirtland, ASSWP Sandia, Holloman AFB, AEC Los Alamos Security, and FBI Albuquerque/El Paso. The full mid-century AF/AEC/FBI New Mexico nuclear-installation security loop, in primary form.