ARRB Electronic Records at NARA

[First posted on December 11, 2017, at]

In all the news coverage about the release of documents under the JFK Records Act, one release has attracted almost no attention. As the NARA press release on the October 26 release noted (here), the electronic records of the Assassination Record Review Board, the agency which determined what government records would be released under the JFKRA, were also released.

The ARRB records are themselves a noteworthy release. Previously few records from the ARRB were available on line, except for their Final Report, and transcripts of the public hearings they held from 1994 to 1998. The Mary Ferrell Foundation website,, has only one set of files from an ARRB staffer (here), and a series of memos from ARRB staffer Douglas Horne (here). The newly released electronic ARRB records are therefore useful indeed for anyone who wants to know not just what the ARRB did, but how and why they did it. The rest of this post provides a summary of the content of the files, as I did earlier for the JFKRA releases (more boring numbers, little on file content).

Number of Files

The ARRB electronic records were released in 83 files. These consist of 13 email archives in csv format, and 70 zip files (all available here). This post will cover only the zip files. The follow-up post will cover the email archives.

In the October release, there are 58 zip files associated with specific ARRB staff members, and another 12 zip files which are general board files. Each zip file is divided into two main directories, Electronic Records and Technical Documentation. The TD directories all have the same components: a readme file explaining a dating discrepancy in some files, and a csv file of metadata for the original files in the ER directory. The ER directory has all the documents from each staff account, converted from their original formats into pdf files, and arranged under sub-directories such as wp-docs, excel, etc.

The pdf files were all created between March and September 2017, the majority in June and July. These of course were not the dates the original files were created, but fortunately the TD directory csv files have the ‘date last modified’ for the original files. All you have to do is match up the directory and file names in the csv file to the pdfs in the zip file.

Oddly, this is not entirely possible, even though the csv files were mostly made in August 2017, after the pdf files were created. There are several dozen files in the zip archives that do not match up with the listings in the csv files. Some of these mismatches are due to things like missing file extensions, files in different directories, file extensions missing or misshapen, misspellings, and some cases that defy explanations, such as becoming There are also pdfs in the zip archives that clearly were not registered in the csv file, and files listed in the csv data that are not present in the zip archives. The numbers are small; files from the zip archives that I have not been able to match back to the csv file lists total 31. pdfs listed in the csv file that do not match to pdfs from the zip archives number only 49. So not a tremendous deal, but I don’t understand how these differences came about.

In addition, a significant number of files from the zip archives are not pdfs of the original file, but instead have been replaced by forms that state the original files were “withdrawn.” These replacement forms are clearly marked by adding the string “_wd_NNNNNNNN” to the file name (wd standing for ‘withdrawn’ I presume). Instead of the original file, these pdf forms consist of a one page document explaining the reasons for the withdrawal.

There are two reasons for file withdrawal: 1) The file “contains electronic characters that are unintelligible and therefore cannot be authoritatively reviewed according to the John F. Kennedy Records Collection Act of 1992.” These files are almost all binary files, with extensions such as .bmp, .dll, .exe, etc. and are thus not worth looking at anyway. 2) The file “has been withdrawn according to (one of the exemptions of) the John F. Kennedy Records Collection Act of 1992.” The file then lists which of the five exemption categories the document falls into (some of them fall into multiple categories). The binary junk files I will call WDI files, the exempt files I will call WDE files. The totals for the different types of files in the zip files is as follows:

File type No. files
Non pdf files 142
pdf files (not withdrawn) 14167
withdrawn pdf files (unintelligible) 2063
withdrawn pdf files (exempt) 396
Total 16768

The non-pdf files are all from the Technical Documentation directories except for two miscellaneous items. The NARA press release states that the ARRB files consisted of “16,627 files from the ARRB drives.” This must refer to the total number of pdf files (and exclude the 140 csv and readme.txt files in the Technical Documentation directories). My count is one more than theirs, who knows what the extra file is (perhaps the novelty mpg file staffer Carrie Fletcher had on her drive).

The withdrawal of files must have been done by NARA, though why they did it according to the JFKRA rules is not clear to me (the JFKRA does not seem to say that it applies to ARRB records), and why the NARA is entitled to carry out such a withdrawal is also unclear. Unlike documents released under the JKFRA, the ARRB withdrawal sheets state that one may apply for the release of the withheld materials under FOIA provisions.

The zip with by far the largest number of files is, which is general “administrative” matters and has 5838 files. After this jumbo zip, the top three zips are from ARRB staff members T. Jeremy Gunn, Laura Denk, and Tracy Shycoff, each of whom has a zip file with from 1000 to 800 plus files. As we should expect, the zips with the smallest number of files are from the interns, and the Board members, who have only two files apiece, except for Board Chairman John Tunheim, who has nine.

I have put up a list of the 58 ARRB personnel with zip files here.

Pages in files

The number of pages in each file ranges from a 2982 page monster (S-ADM-G/PRG/FBIFIX.TXT.pdf; this seems to be a database table dump), to one page scraps. Here is a table of number of pages per file (this does not include non-pdf files or the WDI and WDE files):

Page range No. of Docs
> 100 78
51 – 100 188
31 – 50 168
21 – 30 141
16 – 20 177
11 – 15 297
6 – 10 751
4 – 5 1076
3 1312
2 3978
1 6002

Based on this count, the total number of pages released from ARRB files is 79962.

Dates of files

After matching up the pdfs in the zip files with the files in the csv lists, I can also summarize when the files were last modified:

Year Number of documents
< 1994 63
1994 235
1996 3769
1996 3609
1997 3648
1998 2841

Because the csv lists include the date last modified for the withdrawn files, I could have put these into the summary, but I have chosen to omit withdrawn files from this count, for consistency with the counts above. I have also included in the pre-1994 count unmatched files from the csv lists.

Other than withdrawn files and unmatched files, the pre-1994 records are all text files from various software applications, such as database programs or printer drivers. None of these materials were created for or by the ARRB.

The 1994 files also include a number of such files. The earliest files that clearly were created by ARRB staffers are from September 1994 (by then executive director David Marwell), so that the actual figure for 1994 is more like 189 files.

Most of the staff accounts were set up in January 1995, as dated by the creation of an EXCEL subdirectory with a text file at this time.

The ARRB closed on September 30, 1998, and the last files in this release are some thank you notes by the ARRB’s final executive director Laura Denk, dated September 26.

(There are also three files dated from 2015 whose date stamps I cannot explain. Two of these are excerpts from a 1998 article by David Mantik, the other what looks like a version of a 1995 staff memo)

Withdrawal of files

As the withdrawal notifications provided in the ARRB files indicate, files were reviewed for withdrawal from December 2016 to July 2017, with most of the reviews done in June and July 2017.

There were two reasons for withdrawal of files, as noted above: files had unintelligible (non-ASCII) characters that made review impossible (WDI), or they fell under one of the ‘five exemptions’ of the JFKRA (WDE). In fact, as the withdrawal sheets show, there is one more exemption, 11(A), which exempts tax returns from the JFKRA. This exemption was invoked in four records.

Exemption 6(5), which exempts records revealing Secret Service protection measures, is in fact never invoked at all for the ARRB file withdrawals. The most commonly invoked exemption was 6(3), which was invoked 293 times. This exemption covers records whose release would constitute an unwarranted invasion of privacy. Judging from the title of these records, many were documents dealing with ARRB budget information, and probably revealed Board and staff salaries.

Final notes

There are some quite interesting memos in this release, and overall one gets a good idea of the ARRB’s sometimes contradictory goals and attitudes. My own interest at this point is in comparing what ARRB hoped to release, and what has come out so far. This will be the focus of later posts.