[1996 spin control packet provided to U.S. Attorneys on the Busey Roll Call remarks.] DEPARTMENT OF THE TREASURY BUREAU OF ALCOHOL, TOBACCO AND FIREARMS 200 Granby Street, Suite 339 Norfolk, Virginia 23510 February 28, 1996 MEMORANDUM TO: United States Attorney's Office Norfolk, Virginia FROM: Resident Agent in Charge Norfolk, Virginia Field Office SUBJECT: Potential Brady Issue in ATF Cases Attached for your information is a Department of Justice memorandum dated February 16, 1996, addressed to all United States Attorney's. Also attached is a transcript referencing a roll call training session given by National Firearms Act Branch, Chief Tom Busey on October 18, 1995, a fact sheet referencing the above mentioned roll call training and other related documentation. It should be noted that the above mentioned roll call training transcript may have already been released to the public, pursuant to the Freedom of Information Act. Should you have questions concerning these documents or if I can be of further assistance please do not hesitate to contact me at (804) 441-3190. Glenn N. Anderson Attachment [new document] U. S. Department of Justice Criminal Division Washington D.C 20530 February 16, 1996 MEMORANDUM TO: ALL UNITED STATES ATTORNEYS FROM: JOHN C. KEENEY Acting Assistant Attorney General Criminal Division SUBJECT: Potential Brady Issue in ATF Cases On January 30, 1996, the Bureau of Alcohol, Tobacco, and Firearms (ATF) informed the Criminal Division of statements made by an ATF official concerning the National Firearms Registration and Transfer Record (NFRTR) that relate to cases brought under the National Firearms Act (NFA), 26 U.S.C.  5801, et seq. During a training session for inspectors and other investigative employees at ATF headquarters on October 18, 1995, Tom Busey, then chief of the National Firearms Act (NFA) Branch, made statements to the effect that ATF witnesses typically testify that the NFRTR is 100% accurate when it is not. The transcript of this training session is the subject of a Freedom of Information Act (FOIA) request made by an attorney specializing in the defense of firearms prosecutions and will be released shortly. Although Mr. Busey's comments, by themselves, appear exculpatory, the information and context provided to the Department of Justice by ATF contradict his statements and demonstrate that they are unlikely to affect the integrity of past or pending cases. In particular, we have been advised by ATF that Mr. Busey's statements are based on faulty premises and do not accurately reflect the testimony presented by ATF officials. Further, ATF has assured us that: no erroneous certifications of NFRTR searches have been entered into evidence; no erroneous testimony concerning the NFRTR has been given by NFA Branch specialists; and Mr. Busey's opinions and conclusions, when put in context and perspective, do not undermine the integrity of the NFRTR or affect the ability of ATF to certify the accuracy of the registration status of an NFA weapon. - 2 - According to ATF, given the various search methods used for each lookup, the required supervisory review and analysis of the search record, and the statutory maintenance of records provision of 26 U.S.C.  5841(e), which require the firearm owner to maintain and produce on request proof of registration, there is virtually no chance that NFRTR database errors would result in erroneous prosecution. In light of the representations made by ATF, it is unlikely that Mr. Busey's remarks constitute material required to be produced under Brady v. Maryland, 373 U.S. 83 (1963), and its progeny. Nevertheless, out of an abundance of caution and consistent with the role of the Department of Justice, the Criminal Division advises that you disclose Mr. Busey's remarks to defendants in potentially affected cases. The transcript of those remarks is enclosed. Attached to the transcript of Mr. Busey's remarks is a two page affidavit of Gary Schaible explaining misstatements in the transcript. Both the transcript and the affidavit are being produced in response to the above referenced FOIA request. In addition, the following documents provided to the Criminal Division by ATF also accompany this memorandum: 1. a handwritten sworn statement of Tom Busey dated November 30, 1995; 2. memoranda dated December 1, 1995 and December 11, 1995 and incident report concerning the ATF internal investigation of Mr. Busey's statements; 3. minutes of a meeting held on November 9-10, 1994 to address firearms and explosives data integration; 4. a memorandum dated February 9, 1996 and supporting material constituting the report of the recent audit of the NFA database; 5. a memorandum dated April 30, 1991 concerning the accuracy of the NFRTR; 6. correspondence with Senator McClure and Senator Bayh dating from November 1979 through January 1980 relative to the accuracy of the NFRTR. The Criminal Division advises that you disclose this material in its entirety in order to put Mr. Busey's statements in context. Finally, also enclosed are a legal memorandum discussing Brady far your guidance, as well as a Fact Sheet prepared by ATF that provides background and is intended for your use in evaluating or responding to the potential Brady issue. - 3 - We strongly suggest that disclosure of this information be made in all cases in which a defendant was convicted -- either by guilty plea or after trial -- of a violation of the NFA from 1992 to the present. ATF Headquarters is in the process of identifying all such cases. However, as an initial matter you should also do so, through PROMIS and other case records systems and from records maintained by your local ATF office. Please notify the Criminal Division at the number specified below as to all affected ATF cases in your district. Once you identify these cases in your district you should notify each defense counsel of record of the information recently learned by the Criminal Division and enclose the relevant material. A sample letter to counsel is enclosed for your convenience. In those instances in which defense counsel cannot be notified (e,g., because of counsel's death, retirement or change of address), you should consider notifying the chief district court judge and federal public defender of the information concerning the NFRTR. If you have any questions or need clarification, please contact Ronnie L. Edelman, Senior Trial Attorney in the Criminal Division, at (202) 616-2505. [new document] Date, 1996 John Doe, Esq. Doe & Doe, P.C. 1 Main Street Anycity, USA 12345 Re: U.S. v. Yourclient (no. 94-1234) Dear Mr. Doe: As you may recall, the indictment [or other charging document] in the above-captioned matter contained one or more charges under 26 U.S.C.  5861, relating to the registration of firearms. The Department of Justice recently has learned of material that may be relevant to that case. In particular, enclosed is a transcript of remarks made by Tom Busey, the then chief of the National Firearms Act Branch of the ATF, during an October 18, 1995, training session for ATF inspectors and other investigative employees. During these remarks, Mr. Busey made statements about the accuracy of the National Firearms Registration and Transfer Record ("NFRTR") and of ATF witness testimony concerning the NFRTR. In addition, we have enclosed other documents that relate to Mr. Busey's statements. By providing these materials, the Government does not concede that either the statements made by Mr. Busey or the other materials are favorable to the defendant or material to the defendant's case. Rather, we are doing so in order to avoid any suggestion that the Department has not provided all relevant material in this matter. Please contact me if you believe that we need to discuss this issue further. Very truly yours, JANE JONES UNITED STATES ATTORNEY By: Michael A. Smith Assistant U.S. Attorney [new document] U. S. Department of Justice Criminal Division Washington, D.C. 20530 MEMORANDUM SUBJECT: Some Principles Governing the Brady Analysis Governing Principles. A defendant is entitled to evidence in the hands of prosecutors under Brady v. Maryland, 373 U.S. 83 (1963), if that evidence is favorable to him and is "material" to guilt or punishment. United States v. Bagley, 473 U.S. 667 (1985). In order to be favorable to the defendant, evidence does not have to point directly to innocence; it is enough if the evidence does no more than demonstrate that a factor that "could link the defendant to the crime do[es] not." Patler v. Slayton, 503 F.2d 472 (4th Cir. 1974). Nor need the evidence defeat this factor conclusively; it need only tend to do so. Evidence is "material" for Brady purposes "only if there is a reasonable probability that, had the evidence been disclosed to the defense, the result of the proceeding would have been different." Bagley, 473 U.S. at 682. The "mere possibility" that undisclosed evidence might have affected the trial outcome does not establish materiality far Brady purposes. United States v. Agurs, 427 U.S. 97, 109-110 (1976). As a general matter, the courts find that inadmissible evidence is not material for Brady purposes. See, e.g., Wood v. Bartholomew, 116 S. Ct. 7 (1995); United States v. Kennedy, 890 F.2d 1056, 1059 (9th Cir. 1989), cert. denied, 494 U.S. 1008 (1990); United States v. Ranney, 719 F.2d 1183, 1190 (1st Cir. 1983). However, inadmissible evidence may qualify as Brady if it could lead to witnesses or other evidence that would be admissible. See, e.g., United States v. Phillip, 948 F.2d 241, 249 (6th Cir. 1991); Sellers v. Estelle, 651 F.2d 1074, 1077 n. 6 (1981), cert. denied, 455 U.S. 927 (1982). In addition to purely exculpatory evidence, a defendant is entitled to disclosure of information that could be used to impeach government witnesses. Giglio v. United States, 405 U.S. 150, 154-155 (1972). The same "reasonable probability" standard of materiality applies with respect to impeachment evidence as to other types of evidence. See 2 LaFave, Criminal Procedure  13.5, 184 (1991 Supp.). - 2 - Generally, the good or bad faith of the prosecutor in withholding exculpatory evidence is irrelevant to the Brady analysis. in other words, a prosecutor's failure to produce all material evidence favorable to the defense constitutes a Brady violation even if prosecutor acted in good faith -- that is under the mistaken belief that the evidence in question was not exculpatory. Conversely, withholding evidence in bad faith does not require reversal if the evidence is not exculpatory. See Agurs, 427 U.S. at 110. However, a Court might consider the prosecutor's bad faith in determining whether or not withheld evidence is material on the theory that "[]the fact that the government seeks in bad faith to suppress certain evidence indicates that such evidence may indeed be material." United States v. Jackson, 780 F.2d 1305, 1311 n. 4 (7th Cir. 1986). It is well settled that a prosecutor's constitutional obligation is not violated, notwithstanding the nondisclosure of apparently exculpatory evidence, where the existence and possibly exculpatory content of the evidence is available or known to the defense. See, e.g., United States v. Valera, 845 F.2d 923 (11th Cir. 1988), cert. denied, 490 U.S. 1046 (1989); United States v. Young, 618 F.2d 1281 (8th Cir. 1980), cert. denied, 449 U.S. 844 (1980); 2 LaFave, supra,  19.5, 546 (1984). Application of Principles to Pending Cases. As practical matter, the Brady standard of materiality is difficult to apply at the pretrial or early trial stages because, at those stages, prosecutors cannot know how the trial will play out and thus whether a particular item of evidence is likely to affect the final outcome. For this reason, prosecutors, in the name of prudence, ordinarily turn over questionably material evidence that is favorable to the defense in order to avoid jeopardizing their cases (and being cited for misconduct). In other words, the best Brady policy is to resolve all doubts concerning materiality in favor of disclosure. See Kyles v. Whitley, 115 S.Ct. 1555, 1568 (1995). At the very least, a prosecutor should ask the court to review the evidence in camera to determine if it should be disclosed, instead of taking it on himself to suppress evidence that might qualify as Brady material. See United States v. Schwimmer, 649 F. Supp. 544, 549 n.5 (E.D.N.Y. 1986). Application of Principles to Closed Cases. An already convicted defendant may challenge his conviction on Brady grounds through a motion for a new trial under Fed. R. Crim. P. 33 or a collateral attack on his conviction under 28 U.S.C. 2255. With the full trial record in hand, the prosecution, at the post-trial stage, is in a position to resist such claims on materiality grounds -- that is, by demonstrating that there is no reasonable probability that the withheld evidence would have affected the verdict. This could be accomplished, for example, by a showing that the Brady information consisted of misstatements or otherwise lacked merit, or that the evidence of the defendant's guilt was so strong that this information, even if credited by the jury, could not have made a difference in the final result. - 3 - Courts making a post-trial determination of materiality commonly turn the evidence over to the defense to provide an opportunity to show that the evidence could have been used to its benefit. See 2 LaFave supra, at 185 (1991 Supp.). Under appropriate circumstances however -- as where the evidence is arguably confidential or privileged -- the determination of materiality may be made in camera. See Pennsylvania v. Ritchie, 480 U.S 39 (1987). Because a defendant has two years after final judgment to seek a new trial based on newly discovered evidence, see Rule 33, our obligation under the Due Process Clause to disclose Brady material might arguably extend to closed cases that reached final judgment within two years before the material came into the Government's possession, assuming of course that it would have been material. Moreover, the courts might find that we have a duty to notify defendants of our failure to make post-final judgment disclosure within the two-year period, since defendants might have recourse to Section 2255 for relief from any resulting constitutional violation. With respect to cases that reached final judgment outside the two-year period, a defendant has no judicial recourse for obtaining a new trial based on newly discovered evidence, unless the new evidence establishes a violation of the defendant's constitutional rights (as would for example, evidence of knowing use of perjured testimony by the prosecution). See Guinan v. United States, 6 F.3d 468, 470-411 (7th Cir. 1993). [new document] FACT SHEET FOR UNITED STATES ATTORNEYS TRANSCRIPT OF 10/18/95 ROLL CALL TRAINING GIVEN BY THOMAS BUSEY, CHIEF, NATIONAL FIREARMS ACT BRANCH Attached is a transcript of a roll call training presentation by the Chief, National Firearms Act Branch, Bureau of Alcohol, Tobacco and Firearms (ATF), on October 18, 1995. The transcript includes allegations of errors in the firearms registration records maintained by ATF under the National Firearms Act [NFA) and may be sought by the defense for impeachment purposes. The transcript is also being released to a criminal defense attorney, James Jeffries, pursuant to the Freedom of Information Act. We are forwarding the transcript for your use and information. These documents will be discussed in detail below. I. BACKGROUND A. The National Firearms Registration and Transfer Record The National Firearms Registration and Transfer Record (NFRTR) is the central registry of firearms established by 26 U.S.C.  5841 and consists of the identification of the firearm, identification and address of the person entitled to possession of the firearm, and the date of registration. The National Firearms Act (NFA) Branch utilizes an automated database to retrieve information from the NFRTR. The forms which comprise the NFRTR are microfilmed prior to storage for easier retrieval. B. Searches of the NFRTR Searches of the NFRTR are done primarily by serial number and name. Due to the limitations of the system, the program identifies only records which are an exact match with what is requested. When a serial number search is requested, NFA Branch specialists can only search the NFRTR for the serial - 2 - number that the agent requests. If there is a prefix or suffix to a serial number or one (or more) of the digits is incorrect, then the search will yield negative results. For a name search, the NFA Branch has instituted procedures for searching the system as thoroughly as possible. Ideally, the NFA specialist conducting the search would enter the complete name and conclude that the results are either negative or positive. If the specialist entered the complete name, any typographical errors in the name field of the database would result in the system not identifying the record. Accordingly, the specialist enters only the first three letters of the last name and performs the search. Any record beginning with those three letters will be identified. [footnote 1]If there are any data entry errors in the first three letters of the last name, or if the requester provides a spelling for the name which does not match the first three letters in the name as it appears in the database, this method will not identify the record. If any records are identified with the same last name, but only a first initial, the specialist examines the registration document (on microfilm) to determine whether it could be for the same individual or whether it could be eliminated based on the date of birth or other identifying information provided by the agent. The specialist will also perform another search, limiting the search with the first three letters or the last name to the state of residence of the suspect. [footnote 2] This is done to check the results of the first search and to decrease the number of records which must be examined for a possible ------------------ 1. For example, if the name "Smith" is searched, the program will provide responses identifying every record which begins with "Smith," beginning with those with a blank space after the letter "h," such as "Smith Enterprises," through those with punctuation in the space after the letter "h," such as "Smith, Andrew," and ending with the letter "z" in the space following the letter "h," such as "Smithz." 2. The system is not programmed to perform a search limited to a particular address, street, or city. The only geographical area that can be searched in conjunction with a particular name or serial number is a state. - 3 - match. This lessens the possibility of overlooking a name similar to that being searched which may have been incorrectly entered into the NFRTR. In instances where a common name, such as Jones or Smith, is to be searched, the use of only three letters will identify too many records. To reduce the number of records the specialist would search using the first four letters of the name. After an NFA Branch examiner or specialist completes a search of the NFRTR, a log sheet for the search is forwarded to the branch chief for review. The purpose of the review is to give the branch chief an opportunity to make sure that all established procedures have been followed and that a thorough search of the system has been done. After the branch chief reviews the results of the search, the results are forwarded to the requester. C. Correction of Data Entry Errors in Database ATF is currently in the process of examining the entries in the NFRTR database for the purpose of correcting all data entry errors. ATF personnel are comparing all entries in the NFRTR with a copy of the registration form and will correct any data entry errors that are discovered. This process will be completed in the near future. D. Previous Litigation Concerning Accuracy of NFRTR Issues concerning the accuracy of the NFRTR have been addressed by Federal courts in the past. United States v. Seven Miscellaneous Firearms, 503 F.Supp. 565 (D.D.C. 1980), involved a civil forfeiture of NFA firearms seized from the National Rifle Association (NRA) museum. The forfeiture was based solely on the proposition that the Government had no record that the seized firearms were registered in accordance with the NFA. The NRA had obtained, during discovery, documents relating to a 1975 internal study of the NFRTR citing a number of problems and inaccuracies in the manual record system being used at that time. The NRA used the study to support its argument that the Government could not meet its burden in proving that the seized firearms were not registered. The court held that the items seized were not firearms within the meaning of the NFA and, therefore, were not required to be registered. Although the court did not rule on the adequacy of the NFRTR, the court commented that - 4 - "[c]onsiderable evidence was received that the Bureau's officials have for many years recognized the inadequacy and incompleteness of the Bureau's records. The Court is not required to pass judgment on this, however, because the Government has failed to show that these seven items are firearms." 503 F.Supp. at 578. Criticism of the NFRTR surfaced again in United States v. Stout, 667 F.2d 1347 (11th Cir. 1982). The defendants in this case sought to introduce six internal ATF memoranda. The five oldest documents, dated May, 1375-April, 1976, discussed problems with the mechanical filing system then used for the NFRTR. The sixth memorandum, dated August 20, 1979, restated the problems with the mechanical filing system and explained how the problems had been corrected by manual filing. The defense also sought to introduce an authenticated copy of 1979 Senate Oversight hearings on ATF which included copies of the 1975 and 1976 memoranda but not the 1979 memorandum. The court refused to allow the documents to be admitted into evidence and the defense argued, on appeal, that this was reversible error. The Eleventh Circuit held that the trial court's refusal to admit the exhibits was not an abuse of discretion since the documents were of dubious probative value. The court noted that the 1975 and 1976 memoranda discussed conditions existing 5-6 years before the trial which had largely been corrected. The court also noted that the exhibits were largely hearsay which contained extensive irrelevant material. II. Roll Call Training October 18, 1995 - Tom Busey, Chief, National Firearms Act Branch On October 18, 1995, the ATF Training Division gave a role call training session for all Special Agents and Inspectors at Bureau headquarters. As part of that session, Tom Busey, then Chief of the NFA Branch [footnote 3] gave an overview of the work --------------------------- 3. Mr. Busey requested reassignment in January, 1996, and is now a specialist in the Wine and Beer Branch, Alcohol and Tobacco Programs Division, ATF. - 5 - done by that branch. The session was broadcast on closed circuit television throughout the building. The session was also videotaped by the Training Division [footnote 4]. The following comments are provided concerning statements made during the training which could be damaging to the Government's case. Numbers in boldface correspond to pages and line numbers in the transcript. 1. (3-10) "There's approximately 728,000 Title 2 weapons [registered in the NFRTR] *** (3-18) Of that 728,000, we estimate, because we don't have the time nor the inclination to do it on a monthly basis, anywhere between 150 to 155,000 is the flash grenades. They come in and out of the inventory so quickly, and probably the accuracy of those is not very good, basically because when police departments and other law enforcement agencies use these flash grenades, they're supposed to report to us. *** (4-4) We remove them from the inventory. But it's such a continual turnover. *** Someday when we have the manpower and we have the time, we need to go through and separate these out. COMMENT: Flash grenades are registered in the NFRTR as destructive devices. They are explosive, concussion producing hand grenades that produce a blinding light, a loud noise, and smoke. Flash grenades are commonly used by Federal, State, and local law enforcement agencies in hostage and rescue situations, for dynamic entries, and in training. It is common for police departments to acquire a number of registered flash grenades that are used shortly after receipt. Some agencies report their destruction of these devices on a regular basis and an "expended" entry is made in the NFRTR. They are not removed from the NFRTR, as Mr. Busey states. Other agencies rarely, if ever, report their use of the devices. The fact that the NFRTR may include a significant number of flash grenades that no longer exist does not mean the registry is inaccurate. The system was never intended to reflect transactions, including destruction, that occur --------------------- 4. Roll call training sessions are generally videotaped for use in training field personnel and headquarters personnel who were unable to attend the session. However, due to the inaccurate and misleading statements made by Mr. Busey during this training Session, the Training Division, ATF, has determined that the videotape will not be used to conduct additional training of ATF personnel. - 6 - without notification to ATF. Thus, even after the destruction of the device, it is entirely appropriate for the registry to reflect that the grenades are registered to the specified law enforcement agency. 2. (9-2) "Let me say that when we testify in court, we testify that the data base is 100 percent accurate. That's what we testify to, and we will always testify to that. As you probably well know, that may not be 100 percent true. If our data base was absolutely error free, we could simply run the name of the individual and his first name, and if it didn't come up we could guarantee everyone that that individual doesn't have a Title II weapon registered to him." COMMENT: Mr. Busey has given a sworn statement to the Office of inspection, ATF, (included in the documents which may be subject to Brady) that this statement was a misstatement of the facts. Mr. Busey did not mean to convey the impression that NFA specialists testify that the database is 100 percent accurate. He meant to convey that the database contains certain inaccuracies which are due to human error. Gary Schaible, a specialist in the NFA Branch for over 15 years, states that, to the best of his knowledge, neither he nor any other NFA Branch personnel have ever testified that the data base is 100 percent accurate. (See "Corrections by Gary N. Schaible Concerning Transcript of Roll Call Training" attached to the transcript.) Moreover, no ATF employee has ever testified in court that there was no weapon registered to a defendant when, in fact there was, and no erroneous certifications have ever been introduced into evidence in NFA criminal cases. Certifications prepared by the NFA Branch state that a "diligent search" was conducted, not that there are no errors in the system. 3. (13-13) "what we've started, since there was a problem in Baltimore with a look up and there was a problem up in Minnesota, I think it was about six months ago, from now on, before negative information is sent to an agent -- if the agent indicates that it's routine, he's not in a big rush for it, we used to get it back to him on the same business day." COMMENT: The "problems" referred to are lookups when the NFA specialist advised a Special Agent that there were no weapons registered to the suspect, when in fact, there were. - 7 - Both these incidents occurred before the Chief, NFA Branch, began reviewing the results of NFRTR lookups and resulted in the NFA Branch requiring such review prior to reporting negative search results to a requesting agent. The incident in Baltimore occurred on February 27, 1995, and involved a request from a Special Agent for any firearms registered to "John Ferraro," full name unknown date of birth unknown [footnote 5]. The specialist queried the NFRTR using the letters FER and FET, reviewed 26 entries, and concluded that there were no firearms registered under the name provided. This information was provided to the requester on the same date. On March 2, 1995, a search warrant was executed at the residence of the suspect based, in part, on the results of the NFRTR search. The suspect produced approved registration forms for both firearms which were the subject of the warrant. The next day the Special Agent contacted the NFA Branch with the full name of the suspect, John Armando Ferrari, provided the date of birth, and the serial numbers of the firearms. The NFA specialist performed a search of the NFRTR using the serial numbers provided and identified two registrations to Mr. Ferrari, one under the name of John Armando Ferrari and one under the name of John Armando Ferari. The second registration accurately reflects the name as written on the Form 4. Although the suspect's name was not correctly spelled when the Special Agent first contacted the NFA Branch in February, 1995, the two registrations were included in the 26 entries recovered using the letters FER. However the specialist failed to recognize these entries as a material due to the differences in spelling between "Ferraro" --------------------- 5. The names "John Ferraro," "John Ferrari," "Gary Gwaltney," "Gary Gawaltney," and "Gary Gwantney," are not the actual names of the suspects involved in this lookup and the lookup related to the incident in Minnesota, discussed later in this fact sheet. Disclosure of the actual names of the registrants would violate 26 U.S.C.  6103, which generally prohibits the disclosure of "returns" and "return information". Since the identity of the registered possessor of an NFA firearm is collected in relation to tax liability imposed by the NFA, this information is "return information" as defined in 26 U.S.C.  6103(b) (2) (A). The fictitious names used in the discussion of these incidents do not appear in the NFRTR and were chosen to demonstrate the data entry errors and other human errors that actually occurred. - 8 - and the two names shown on the registrations. Thus, this incident was the result of an error in judgment rather than incorrect data entry. Mr. Busey's reference to an incident in Minnesota concerned a request for a lookup by a special Agent in St. Paul. On January 18, 1995, the agent contacted the NFA Branch and asked that a search of the NFRTR be done to determine whether Gary Gwaltney of Minneapolis, Minnesota, had a silencer registered to him. The NFA specialist who took the call searched the NFRTR for weapons registered to "Gwaltney, Gary," using the first three letters of the last name, "GWA." The specialist advised the Special Agent that Mr. Gwaltney had a machinegun registered to him, but no silencer. The same day, the agent interviewed Mr. Gwaltney who stated that he had properly registered the silencer with ATF and would bring a copy of the approved Form 4 to the agent's office the next day. The agent again contacted the NFA Branch on January 19, 1995, and requested that another lookup on Mr. Gwaltney be done. The specialist who did the lookup searched the NFRTR for weapons registered to "Gawaltney, Gary John," using the first three letters of the last name, "GAW." It is not clear whether the specialist made an error in writing down the name given by the Special Agent nor whether the Special Agent misspelled Gwaltney's name. This search disclosed no weapons registered under the name of Gary John Gawaltney, and the results were given to the requesting agent. Later in the day on January 19, 1995, Mr. Gwaltney cave the Special Agent copies of his Form 4 registrations for the silencer and the machinegun. On January 20, 1995, the Special Agent contacted the NFA Branch and asked hat a search of the NFRTR be done far any registrations for the silencer and machinegun listed on the Forms 4, and provided the model and serial numbers of the weapons. The search disclosed that the machinegun was registered to Gary John Gwaltney and the silencer was registered to Gary Gwantney. The specialist reviewed the registration form for the silencer, which indicated that the registered owner's name was Gary Gwaltney. The specialist corrected the data entry error in the database, changing "Gwantney" to "Gwaltney." The specialist advised the Special Agent that both weapons were properly registered to Mr. Gwaltney. This incident demonstrates the problems presented in searching the current system. The specialist who did the initial search used the correct spelling of the suspect's - 9 - name and did a search using "GWA." Even though the registration for the silencer was under the name "Gwantney," the specialist should have reviewed this registration in the search results, recognized the similarity in the names, and referred this information to the Special Agent. The second search done on January 19, 1995, under the name "Gawaltney" could not have identified the registration under "Gwaltney," or the registration under "Gwantney," since search of the letters "GAW" would not have retrieved either registration. 4. (16-4) "So we're hoping that eliminates the possibility that anything goes out erroneous, because we know you're basing your warrants on it, you're basing your entries on it and you certainly don't want a Form 4 waved in your face when you go in there to show that the guy does have a legally registered Title II weapon. I've heard that's happened. I'm not sure." COMMENT: An approved Form 4, Application for Transfer and Registration of Firearm, is returned to an applicant (the transferor) when the transfer of a firearm is approved. The transferee is required to retain the approved form as evidence that his firearm is properly registered in the NFRTR. 26 U.S.C.  5841(e). This reference is probably to the incidents in Baltimore and Minnesota, discussed above. 5. (17-17) "If information that's in the data base is not accurate, it doesn't make any difference how good of a search we do, it'll come out wrong." COMMENT: It is incorrect to conclude that errors in the database will always result in a bad lookup. As stated above, NFA specialists have devised different methods to make their searches as accurate as possible. 6. (17-20) "So the information on the 728,000 weapons that are in the data base has to be 100 percent accurate. Like I told you before, we testify in court and, of course, our certifications testify to that too when we're not physically there to testify, that we are 100 percent accurate. *** (18-4) But we have found instances in our records where names have been misspelled, they've been inverted; vowels "ie" have been changed. And, of course, computer programs only pull up what you put in." - 10 - COMMENT: As stated above, Mr. Busey has provided a sworn statement that his words were not meant to convey the impression that NFA specialists testify that the database is 100 percent accurate. He meant to convey that the database contains certain inaccuracies which are due to human error. 7. (18-9) "We've made monumental strides in correcting this. A major correction event took place in 1986. About a year ago, we instituted a quality review team in the division. That's three individuals who review every transfer record that goes through an examiner to register a Title II weapon, or to transfer a Title II weapon." COMMENT: The "major correction" referred to is the computerization of the NFRTR around 1986. This resulted in quicker, more reliable searches of the registry. A quality review team (QRT) was established in July, 1994, to provide for the error free entry of data into the NFRTR. The team is composed of experienced personnel from the NFA Branch and the Firearms and Explosives Imports Branch who review and correct data entry into the NFRTR. The QRT reviews NFA forms after they have been reviewed and entered into the NFRTR by an examiner but prior to their approval and return to the applicant. The QRT does not review every field on the form or in the database, but focuses on name, city and state of the transferee, and the serial number and description of the firearm to be transferred, when the quality review team determines that an error has been made either in procedural steps or data entry, the team discusses the error with the examiner to ensure the mistake will not be repeated. All errors are corrected before the application is returned to the applicant. 8. (19-3) "This quality review team, when I first came in a year ago, our error rate was between 49 and 50 percent, so you can imagine what the accuracy of the NFRTR could be, if your error rate's 49 to 50 percent. The error rate now is down to below 8 percent, and that's total. That's common errors and critical errors." COMMENT: The reference to an error rate of 49-50 percent is based on an informal estimate conducted in 1994 at the request of the Chief, Firearms and Explosives Regulatory Division. No written reports or other documents concerning this estimate can be found and it is unclear who actually prepared the estimate. The Chief, Firearms and Explosives - 11 - Regulatory Division, advised Mr. Busey that the error rate in the NFRTR was 49-50 percent when he became the Chief of the NFA Branch in September, 1994. In addition, minutes of a meeting concerning problems with the NFRTR held at the ATF Tracing Center on November 9-10, 1994 (copy attached), state that "two months ago an error rate of 50) was found." Officials from Regulatory Enforcement, Criminal Enforcement, and the Information Services Division were present at the meeting. The reference to an error rate of 8 percent is probably based on the error reports prepared by the QRT, which prepared weekly and monthly reports regularly since being established 1994. The September, 1994, report indicates that beginning on or about October 3, 1994, the review team began making a distinction between what they referred to as "common" versus "significant" errors. Generally, the QRT considered "common" errors to be those that would not affect a records search, such as restrictions relating to possession of the firearms and "pending" annotations not being properly recorded in the database. Errors which were considered "significant" were those affecting the accuracy of the search, including misspelled or incomplete names or approval of the wrong firearm to the transferee. However the team also indicated as "significant" errors the payment of the transfer tax not being posted and failure of the examiner to keep one copy of the transfer application for return to the transferee. Eventually, at the direction of the Chief, Firearms and Explosives Regulatory Division, the review team made no distinction between common and significant errors. Beginning with the November, 1995, report, no such distinction was made. The error reports show the number of forms reviewed, the number of errors found, and in most cases. the percentage of error. The October, 1995 report shows a common error rate of 6 percent and a significant error rate of 1 percent. The most recent report for November, 1995, shows a total error rate of 6 percent, with no distinction between common and significant errors. At the request of the Deputy Director, ATF, personnel from the NFA Branch and Audit Services Division recently conducted an audit of the name field in the NFRTR. The purpose of the audit was to determine whether the 50 percent error rate mentioned by Mr. Busey in the training session was accurate. The audit was based on a random sampling of - 12 - 350 entries in the NFRTR. According to the report of the Chief, Audit Services Division, this sample size is more than adequate for the entire database of 982,471 entries in the NFRTR. The names of the registered possessors in the NFRTR database were compared against the names on the corresponding hard copies of the registration forms to determine the number of data entry errors in the first three positions of the name. Since NFA personnel use the first three letters of the name to conduct a search, data entry errors in these positions are critical to an accurate search. Data entry errors beyond the first three letters of the name or in the other fields of the database, e.g., city, state, model, and serial number, should not result in an erroneous lookup if the specialist reviews all the entries retrieved and follows the search procedures discussed above. The audit disclosed no errors in the first three positions of the name for the 350 records examined. Projecting this error rate onto the entire database could result in a maximum error rate of 1.5 percent. The audit also tested the entire name field and identified 13 errors in the 350 entries examined, or a 3.7 percent occurrence rate. Projecting this error rate onto the entire database could result in a minimum error rate of 2.2 percent and a maximum error rate of 6.8 percent. A copy of the audit report and a memorandum explaining the report are attached. [footnote 6] 9. (19-22) "The only way we can go back, we have a project -- we established a project, we established a task force. We haven't begun yet because we haven't converted to the new data base. As soon as the new data base comes into effect we'll begin the task-force assignment." (20-6) "What we're doing to do is we're going to go back, starting with the latest entry and working back to the oldest entry and review every hard copy of every document with its entry into the data base to if it's correct. I think originally we figured this would take 781 man days to -------------------------- 6. One of the documents prepared in support of the data summary is not included since it includes "return information" prohibited from disclosure under 26 U.S.C.  6103. The document is a summary of the 13 errors found in the names of the persons to whom NFA firearms are registered. Since this information would disclose the identity of registrants, it is clearly return information. - 13 - do this with five people sitting at a computer eight hours a day. But it's the only way that we can feel that we can ever get it completely accurate. It was fine to begin putting everything in accurate a year ago or at least be guaranteed a year ago it was accurate, but what are you going to do with the entries that go back to the early '80s and the '70s and the '60s? This is the only way we feel we could correct it." COMMENT: Mr. Busey concludes that the long term answer to the problem of bad lookups is to review every entry in the NFRTR and compare it to the hard copy. This approach is consistent with memoranda prepared by Mr. Busey and other NFA Branch personnel on February 16, 1995, and November 8, 1995. Both these documents discuss a proposal to have ATF Personnel "re-key" all the information in the NFRTR database, starting with the most recent information and working back either to the first entry in 1934 or from 1986 present. The memorandum dated November 1995, discusses a meeting at the Tracing Center on November 7, 1995, at which the Special Agent in Charge of the Tracing Center offered to make equipment and personnel available to the NFA Branch to perform this task. Both memoranda recommend that this project be started as soon as possible. The memorandum dated February 16, 1995, is initialed by the Chief, Firearms and Explosives Regulatory Division, to indicate approval. There is also a handwritten annotation at the bottom of the memorandum advising the Chief NFA Branch, to discuss the project with the other Regulatory Enforcement and Criminal Enforcement Division Chiefs. As stated above, ATF is currently in the process of examining the entries in the NFRTR for the purpose of correcting data entry errors in all fields of the database. ATF personnel are comparing all entries in the NFRTR with a copy of the registration form and will correct all data entry errors. This process will be completed in the near future. Significantly, the memorandum from the office of Inspection forwarding the report on the statements by Mr. Busey at the training session stares, "it is the opinion of this office that the entire NFRTR database is in need of thorough review and correction to ensure its accuracy," The memo further notes that the writer (Assistant Director (Inspection)) is aware that the SAC of the Tracing Center has offered the assistance of his staff in correcting the errors in the NFRTR. Since this memorandum may also be Brady material, we are attaching a copy. - 14 - III CONCLUSION Despite the errors in the NFRTR discussed above, ATF has the ability to accurately certify to the registration status of an NFA weapon. Search procedures have been designed to ensure that every lookup is thorough and reliable. In the vast majority of cases, searches performed by NFA Branch personnel are accurate. Indeed, by the time a certification is entered into evidence or an NFA Branch specialist testifies at trial, there is virtually no possibility that the certification or testimony is inaccurate. As previously stated, no erroneous certification has ever been introduced into evidence nor has any erroneous testimony concerning registration of a firearm been given by an ATF employee.