The paper itself.
The archive cites 3,325 sources. This is where the ones we can lawfully host live — in full, unaltered, with a guide to what is on which page. 6 documents, 33 MB.
We did not obtain any of these; the releasing agency did, and someone else published them first. What we add is the reading: which page carries the sentence a case file quotes. Government file trees move and archives go down — four of the first seven source URLs checked for this project had already moved or started refusing requests. A citation that resolves is worth more than one that used to.
July 1995
Government Records: Results of a Search for Records Concerning the 1947 Crash Near Roswell, New Mexico
U.S. General Accounting Office, report to Rep. Steven H. Schiff. GAO/NSIAD-95-187
The most-cited and least-read document in the Roswell case. A congressman asked the government’s own auditors to go and look for the records; this is what they found, and what they found missing. Twenty-six pages, fully searchable, and it does not say what either side says it says.
What is in it → PDF · 26 pages · 0.3 MB
1997
The Roswell Report: Case Closed
Capt. James McAndrew, Secretary of the Air Force Declassification and Review Team; Headquarters United States Air Force
The Air Force’s second and final report, and the one that tried to explain the bodies. Whatever you make of its conclusions, it is the official position stated at length with its sources attached — 232 pages including signed witness statements and an appendix of dummy-drop locations. It is quoted constantly and read rarely.
What is in it → PDF · 232 pages · 18.7 MB
10 July 2009
Unclassified Report on the President's Surveillance Program
Offices of Inspectors General of the Department of Defense, Department of Justice, Central Intelligence Agency, National Security Agency and Office of the Director of National Intelligence. Report No. 2009-0013-AS
The official account of the warrantless surveillance programme, written by the five inspectors general who had access to the classified record and required by statute to report to Congress. It is the document the case file is built on, and it is short enough to read in an evening.
What is in it → PDF · 43 pages · 2.7 MB
13 March 1962
Justification for US Military Intervention in Cuba (Operation Northwoods)
U.S. Joint Chiefs of Staff
This is the document the case file is about. It is short, it is signed, and it says what it says — which is why the page you are reading exists: so the claim and the paper can be checked against each other in two clicks rather than taken on trust.
What is in it → PDF · 15 pages · 0.8 MB
3 August 1977
Project MKULTRA, The CIA's Program of Research in Behavioral Modification
U.S. Senate, Select Committee on Intelligence and Subcommittee on Health and Scientific Research, 95th Congress, 1st Session
Almost everything publicly known about MKULTRA runs through this one day. It is the hearing at which the CIA, having said the records were destroyed, produced the ones that were not. The transcript is fully searchable in this copy.
What is in it → PDF · 173 pages · 8.8 MB
28 January 2021
Mechanisms of slab avalanche release and impact in the Dyatlov Pass incident in 1959
Johan Gaume and Alexander M. Puzrin, Communications Earth & Environment 2:10 (2021)
The most-cited modern attempt to give the avalanche hypothesis a physical mechanism. It is worth reading in full rather than in summary, because what it claims is narrower than what it is usually reported to claim.
What is in it → PDF · 11 pages · 1.9 MB
What gets hosted
Only two things: US federal government works, which are public domain, and openly-licensed research whose licence permits redistribution. Books, news articles and press-agency material are cited and linked, never hosted.
A document also has to earn the space. A mirror of a PDF that is already easy to find, and that we have nothing to say about, is not worth the bytes — it will not outrank the original and it makes the site heavier. The test is whether we can tell you what is on which page.
Nothing here is re-compressed, re-OCR’d or cleaned up. If a scan is poor, that is a fact about the record.