David B. Kopel [a1]
1999 L. Rev. Mich. St. U. Det. C.L. 171. More by Kopel on the Bureau of Alcohol, Tobacco, Firearms, and Explosives, and BATFE tracing.
Law Review of Michigan State University Detroit College of Law
Spring 1999 Copyright (c) 1998 Detroit College of Law at Michigan State University; David B. Kopel
Introduction .............................................................. 171
I. Traced Guns Are Not a Surrogate for Crime Guns in General ............ 173
A. Only a Small Fraction of Crime Guns Are Traced by BATF ............ 173
B. Most Gun Traces Are Not Associated with Violent Crime ............. 175
C. Traces Do Not Reveal How or Why a Gun Was Moved Interstate ........ 177
II. "Assault Weapons": Police Data Show BATF Traces to Overestimate
Criminal Use by 1000% .............................................. 179
Conclusion ................................................................ 184
In ancient Greece, priests would slaughter a sacrificial animal, and then carefully examine the animal's entrails. The priests and their followers both believed that by "reading entrails," one could forecast the future. This process, fortunately, has become less messy these days: rather than using entrails, our modern fact-inventors use something much cleaner, but no more reliable: trace data from the Bureau of Alcohol, Tobacco and Firearms ("BATF"). Today, powerful leaders also make unsupportable claims based on "information" which was never intended to be used in such a manner.
Almost every major firearms control proposal (including "assault weapons" bans,  the Brady Act , gun purchase rationing,  bans on small handguns , and eliminating home-based firearms dealers ) is touted on the basis of the "scientific" evidence provided by BATF traces. The staff of Representative Charles Schumer (D-N.Y.) earned itself a large amount of favorable publicity in 1997 by using BATF traces to create what Representative Schumer called "the first study that shows conclusively that gun control works."  Unfortunately, BATF tracing data was never intended to be used for policy guidance and is unsuitable for that purpose. To build the case for a particular gun law on the basis of BATF traces is to admit that there is no relevant data to support the law.
Only a small and unrepresentative fraction of guns which are seized by the police are ever traced by BATF and, for this reason, BATF has repeatedly stated that its trace data cannot be used to draw conclusions about patterns of criminal gun use or acquisition. The BATF's National Tracing Center advises police departments that the trace data "ONLY reflects trends relating to those firearms for which a trace request is submitted and is only as accurate as the information provided by the trace requesters."
In contrast to certain other federal programs (such as the FBI's Uniform Crime Reporting system), BATF's tracing program was never designed for sociological data gathering. To the contrary, BATF tracing was designed to provide law enforcement with a quick, easy, low-cost opportunity to trace the history of a particular gun as part of an investigation of a particular crime. But do the BATF traces also provide accurate information about the nature of armed crime in general? 
Part One below explores various reasons why gun traces are not good substitutes for information about actual gun use and sales patterns. Part Two examines one case in which it is possible to compare BATF trace data with actual police crime data, namely, the use of "assault weapons" in crime.
Under federal law, all firearms manufactured or imported into the United States must have a serial number.  Firearms manufacturers, wholesalers, and retailers are required to keep precise records of the serial number of each gun they transfer.  Thus, if a particular gun is found at a crime scene, the BATF can, using the serial number, trace the chain of custody of the gun from its manufacture to its sale at retail. From there, law enforcement authorities can interview the retail buyer, to attempt to investigate how the gun eventually came into criminal hands (e.g., the gun was stolen from the lawful retail purchaser).
In an average year, there are at least a million violent crimes committed with firearms.  Each year, the Federal Bureau of Alcohol, Tobacco and Firearms ("BATF") is asked to trace about 25,000 violent crime guns (plus many more guns associated with possessory offenses), or about less than one gun per forty violent gun crimes. About one in two gun homicides results in a BATF trace, compared to one in fifty gun assaults, and one in a hundred gun robberies. 
One reason that few violent crime guns are traced is that information about the chain of custody from manufacturer to retail sale is often not necessary for prosecution of state and local gun crimes.  After all, a District Attorney bringing an armed robbery case needs to prove that the defendant used a gun, not that the defendant used a gun with a particular pedigree. In some cases, local police may find it faster to conduct a trace themselves than to ask BATF to perform the trace. 
Further, some jurisdictions, such as New York, Maryland, California, New Jersey, and Massachusetts, keep detailed records of all legal ownership of handguns, or of all guns.  These jurisdictions would logically use their own records first for gun tracing, and would turn to BATF only when their own data failed. 
The small percentage of guns selected for a trace request are not a random sample, but rather a select group chosen by local police departments.  According to basic statistics theory, a non-random sample is very unlikely to accurately represent the larger whole. 
Accordingly, the Congressional Research Service cautions that the "firearms (which the) Bureau of Alcohol, Tobacco and Firearms selected for tracing cannot be considered representative of the larger universe of all firearms used by criminals or any subset of that universe," because "the firearms selected for tracing do not constitute a random sample." As a result, "ATF tracing data could be potentially biased." 
One reason that BATF traces are not representative is that BATF currently rejects most requests to trace pre-1990 guns, since data for these guns are less readily available. 
It is easy to see how the pre-1990s exclusion could skew results. In the 1990s, the majority of handgun sales were semiautomatics, while most handgun sales from earlier years were revolvers.  A trace sample limited to post-1990 guns would likely overstate the prevalence of semiautomatics relative to revolvers.
Similarly, the refusal to trace pre-1990 guns will also skew the types of long guns which are traced. So-called "assault weapons" were big sellers in the 1980s and early 1990s, but were a very small part of the firearms market in most prior decades.  Thus, limiting traces to only guns made after 1990 will artificially inflate the percentage of "assault weapons" which are traced.
Finally, studies have found that between forty and sixty percent of BATF gun traces fail.  This creates a situation similar to a pollster finding that half of the persons polled refuse to answer certain questions. Elementary statistics theory requires that a data gatherer not ignore the "non-response bias."  Thus, the subset of guns for which traces are successful is even less likely to resemble crime guns in general.
Violent crimes account for only one-seventh of BATF traces.  At least half of BATF traces are for possession offenses.  To the extent that any generalizations can be drawn from BATF trace data, these generalizations are about gun owners whose gun possession violated some kind of ordinance or statute, but who did not use their gun for illegal violence.
In 1992, anti-gun lobbyists touted BATF trace data from the first nine months of 1991 to argue that as much as forty-one percent of New York City "crime" guns came from Virginia.  But of the New York City firearms traced to Virginia during the first nine months of 1991, only thirty- two guns (or one-sixth of the traces) were used in a violent crime.  The rest were associated with technical violations of New York City's arduous handgun licensing scheme, or other non-violent offenses.  Forty-seven percent of the violations involved weapons possession crimes (including simple possession of an unlicensed gun in the home); thirty-five percent involved other non-violent offenses (such as possessing a handgun and a gram of cocaine in the same apartment). 
A 1970s national analysis of handgun seizures found that twenty to twenty- five percent of police handgun seizures were not associated with any crime, not even a licensing violation.  Some of the guns traced by BATF might just have been turned into the police by lawful owners who wanted to get rid of them. (For example, a widow who wanted to dispose of her husband's hunting rifle). 
In New York City, obtaining a handgun license is very difficult. Although New York law requires the police to act on license applications within six months, delays of nearly a year are routine-even for crime witnesses who are being threatened by criminals out on bail.  In addition, it is nearly impossible for an applicant to get a license to carry a handgun, unless the applicant is named "Donald Trump," in which case the carry permit will be granted in a few days.
Because obtaining a New York City handgun permit is very difficult for unwealthy people who cannot afford lawyers, many citizens obtain handguns illegally; they adopt the adage that "it is better to be judged by twelve than to be carried by six." They would rather face the risk of prosecution for an unlicensed handgun than face the risk of living in New York City without a handgun.
Much the same story can be told for Washington, D.C., where crime is even worse than New York City, where the police are notoriously ineffective, and where handgun purchases and possession are entirely illegal. 
to the extent that any conclusions could be drawn from BATF trace data, the
conclusions show that guns from other states are used to evade a prohibition
(Washington, D.C.) or near-prohibition (New York City) on handgun ownership by
non-elites. The trace data, by themselves, do not show that other states are
the main source of guns for violent criminals in New York City or Washington
D.C., because five-sixths of traces do not involve violent crime.
Indeed, the artificially-created "gun criminals" in New York or Washington D.C., who own handguns under circumstances which would be entirely lawful in Virginia or other states, (possession of a handgun for home protection) are much more likely to own a traceable handgun than are actual violent criminals. A person owning a handgun for home defense would have little reason to file off the serial number. However, according to a study sponsored by the National Institute of Justice, sixty percent of actual felons consider a gun's untraceability to be "very important" and another twenty-one percent consider it to be "a little" or "somewhat" important. 
Of course no one would even bother to attempt to trace a gun from which the serial number had been removed (sometimes the serial number can be removed through advanced forensics, but the process is difficult, and used very rarely). This is one more reason why the traced guns are generally not similar to non-traced guns.
Gun traces show the history of the gun from its manufacture, purchase by a wholesaler, sale to a federally-licensed gun dealer, and then retail sale. Except when special investigators are assigned to perform field interviews, traces do not, and cannot, trace the history of the gun beyond retail sale. Hence, a trace may discover that a gun was manufactured in Connecticut, purchased by a Florida wholesaler, and sold to a customer at a Houston gun store. If police confiscate the gun in New York City, the trace does not tell us whether:
The widely-publicized study by Representative Charles Schumer's staff presumes that the first possibility is the only possibility, since it attributes all interstate gun movement to gun-running, never mentioning other, more common possibilities.  The BATF data which Schumer used does not specify how the guns moved between states.
The tracing reports conducted by BATF do not even investigate how a gun moved from its state of retail sale to the state where the gun was found. For example, the 1991 BATF report on gun traces from New York City to Virginia cautioned, "Project Lead does not attempt to determine if a Virginia gun found in New York had been stolen from a Virginian, and then transported to New York." 
Contrary to the Schumer assumption that interstate gun movement must be the result of organized gun-running, a BATF study of interstate gun transport found "that the majority of firearm movement from (s)tates is occurring on an individual basis. That is to say that an individual will acquire a firearm in another (s)tate through the actual purchase by relatives and friends and then transport the firearm back."  Self-protection is the primary motive for such transfers, the BATF reported. 
Of course, an intra-family transaction might not necessarily be legal; federal law makes it illegal for persons other than licensed dealers to sell handguns to persons from another state. And criminals, like ordinary citizens, do buy guns for self-defense. Nevertheless, the BATF data make it clear that interstate gun transfers are usually small-scale and involve guns sold through private transfers, not from retail outlets.  Thus, the picture painted by gun control advocates such as Representative Schumer-a picture of large volume interstate transfers of guns purchased directly from licensed retail dealers-is contrary to the known data.
The difference between the Schumer scenario and the BATF facts has major policy implications. If the Schumer scenario were true, it would support his proposals for even greater federal regulation of retail gun sales. Federal regulation not only would make sense because licensed dealers, wanting to keep their license, would obey federal gun laws, but also because licensed dealers are (supposedly) the main source of interstate guns.
But in truth, the data shows just the opposite. Interstate transfers are primarily private transfers between family and friends. There is no realistic possibility that these individual transfers can be controlled by federal statute. To the extent that the interstate transfers involve transfers between two family members, both of whom are criminals, it is especially unlikely that federal statutes would have any preventive effect. If the private interstate transfer is somehow discovered by the police, current federal laws already provide severe mandatory sentences for illegal interstate gun sales and for the transfer of firearms to persons with any felony conviction or with misdemeanor domestic violence convictions. 
"To hell with statistical theory," some people might say. "Whatever the objections raised by skeptics, we still think that gun traces provide good data about gun crime and gun sales in general, even if the figures might be a little rough here and there."
If one accepts this argument, then one policy conclusion becomes inescapable: one-handgun-per-month laws are a failure and should be repealed. The laws specifying that citizens may purchase only one handgun per month are based on the premise that purchases of multiple handguns from gun stores are a major source of supply for interstate gun-runners. According to the Schumer study, two of the three states that supplied the most guns to New York were Virginia and South Carolina, the only states in the nation with one-gun-a-month laws at the time of the study. 
But dropping one-gun-a-month laws solely because of the Schumer study (or because of any other analysis of trace statistics) would be a mistake. We know that trace statistics can result in errors of more than 1000%, and therefore, trace statistics are so far removed from real-world experience to be of no value in determining crime policy. (Again, to point out this fact is not to denigrate the BATF tracing program; the program was designed to solve individual crimes, not to gather sociological data). One of the most notable real-world instances in which tracing data was used to "prove" facts which were wholly contrary to reality was in regard to so-called "assault weapons."
"Assault weapon" is a marketing term, whose meaning varies depending on whether the user of the term is a member of the gun industry or a gun control advocate. The term generally refers to firearms that have a military-style appearance. Appearance notwithstanding, "assault weapons" are functionally indistinguishable from normal-looking guns: they fire only one bullet with each press of the trigger and the bullets they fire are intermediate-sized and less powerful than the bullets from big game rifles.
But starting in 1989, Handgun Control, Inc., and other anti-gun organizations, gained national attention by claiming that assault weapons were the "weapon of choice" among criminals. 
The political gun ban campaign was significantly bolstered by an article on BATF traces-a report that, considerably later, was shown to be grossly misleading. In May 1989, two reporters from the Cox newspaper chain conducted a study of BATF firearms traces.  The reporters found that for some crimes, assault weapons were involved in approximately ten percent of the traces.  Since "assault weapons" constitute only about one percent of the total firearms stock (the reporters asserted), the ten percent trace figure indicated that "assault weapons" were disproportionately involved in gun crime. "An assault gun is twenty times more likely to be used in crime than a conventional firearm," Cox newspapers claimed.  Politicians who wanted to ban guns took up the line.  Police data, however, showed the ten percent figure to be false. 
The Cox report gave trace percentages for both the nation as a whole (ten percent) and for selected major cities.  The percentage of "assault weapons" reported by Cox newspapers, based on the BATF traces, was ten percent for Chicago, nineteen percent for Los Angeles, eleven percent for New York City, and thirteen percent for Washington, D.C.  In each of those cities, police departments conducted complete counts of all guns that had been seized (not just the guns for which the police department requested a BATF trace). According to the actual police department counts of crime guns in each city, the percentage of "assault weapons" were only three percent for Chicago, one percent for Los Angeles, one percent for New York City, and zero percent for Washington, D.C.  Thus, when the BATF trace sample was compared with the comprehensive police gun seizure data, BATF traces over- stated the percentage of assault weapons used in crime by over 1000% for Los Angeles, New York, and Washington.
weapons" were a high percentage of BATF gun traces, but a small
percentage of total crime guns.
Sources: BATF; police departments in respective cities.
There are several reasons why "assault weapons" are more likely to be selected for a trace request.  Many "assault weapons" have an unusual appearance, which might raise curiosity (and a trace request) compared to an "old-fashioned" gun, such as a Smith & Wesson .38 Special. 
The publicity surrounding "assault weapons" also may have increased police interest in these weapons, increasing the likelihood that a trace would be requested.  As the Congressional Research Service noted, "a law enforcement officer may initiate a trace request for any reason."  "If . . . law enforcement offices in certain regions have determined that certain types of firearms (such as military-style semiautomatics that accept large capacity magazines) should be traced because they are thought to be used by dangerous offenders, the data in the tracing system will reflect those specific concerns." 
Polling data show that high-ranking police administrators, such as big- city police chiefs, may be far more supportive of gun control in general, and assault weapon prohibition in particular, than are mid-ranking or street-level law enforcement officers.  As a result, for the reasons explained by the Congressional Research Service, heightened policy administration concern about "assault weapons" could result in a disproportionate percentage of such guns being submitted to BATF for tracing.
In addition, almost all "assault weapons" were first sold at retail after 1968. Before 1968, gun retailers who sold only to in-state customers did not need a federal firearms license.  Also, before 1968, federal law did not require firearms dealers to maintain registration records of retail sales.  Thus, successful traces are more likely to be conducted on guns made after 1968-a category which includes a higher percentage of the total stock of assault weapons than of the total stock of, for example, bolt-action rifles. While the Cox study was in progress, BATF had a policy of not accepting trace requests for guns made before the early 1980s. This policy certainly would cause an increase in the percent of traces involving "assault weapons."
It should be noted that the discrepancy between the BATF traces and the actual crime gun seizures was not confined to the four major cities discussed above. Researchers have now obtained comprehensive crime gun data for many cities based on actual inventories of firearms seized by the police, in not one of the cities does the percentage of "assault weapons" seized even remotely approach the BATF trace figure of ten percent. The highest figure was four percent; one percent (or less) was much more common.  Accordingly, it can only be concluded that BATF firearm trace requests are not an accurate mirror of actual firearm use in crime.
To accompany the temporary federal prohibition on new "assault weapons,"  Congress in 1994, ordered the Attorney General to study whether the ban was changing patterns in "assault weapon" use by criminals. When asked if data existed to allow such a study, the FBI's Uniform Crime Reports (UCR) Section replied, "The UCR Section knows of no existing data to provide a basis to address the question."  Stated less delicately, the FBI's statistical experts do not believe that gun traces provide reliable information about the general use of guns in crime. Lobbyists and politicians who attempt to use trace data for this purpose should explain why they think that they are right and the FBI is wrong.
To advocates of restrictive gun laws, BATF traces are like psychoanalyst's inkblots: the viewer always finds in them exactly what he wants to see. Thus, BATF traces are used to create allegedly "scientific" data to promote the banning of small and inexpensive guns,  to demonize certain large handguns,  to prove that the Brady Act is working,  to prove that the federal assault weapons ban is working,  to prove that more controls are needed on "assault weapons,"  to make the case for eliminating firearms dealer licenses for persons who operate small businesses from their home,  and to prove that gun rationing laws work. 
The (mis)use of BATF tracing reports by gun control lobbyists has often worked. The federal "assault weapons" ban, the Virginia gun rationing law, and the federal elimination of home-based businesses from the ranks of licensed firearms dealers all became law with the significant help of claims made about BATF traces. Proposals for further controls, such as federal licensing of gun owners, federal gun purchase rationing, and state and federal bans on small handguns are all predicated on the allegedly scientific information derived from BATF traces.
The English philosopher Jeremy Bentham once derided natural rights as "nonsense on stilts."  Whatever one thinks about natural rights, Bentham's phrase is an apt description of attempted use of BATF traces to support gun laws.
In the 19th century, practitioners of the "science" of phrenology conducted extensive measurements of people's skulls and used the data to produce all sorts of findings, all of which had the appearance of hard science, but none of which turned out to have any validity. For example, persons with certain shapes of skulls, dubbed "low-brow," were said to be intellectually inferior. The link between BATF gun traces and real-world use of violent crime guns is just as weak as the link between skull shapes and individual intelligence. BATF traces and the measurement of skulls both generate a large amount of data, but that data is not scientific evidence upon which criminal justice policy can be based.
Phrenology, unfortunately, fits the prejudices of its era, since it helped "prove" that Blacks and Eastern Europeans were intellectually inferior to people from northwest Europe. Likewise, BATF gun trace studies fit the prejudices of our own time, since they are designed almost exclusively to "prove" the need for more gun laws.
Unlike phrenological measurements, BATF gun traces do have a valid use: tracing the history of a particular gun used in a particular crime, up to the point of retail sale. The trace data should be released from their involuntary enlistment in the cause of pseudoscientific support for restrictive firearms legislation.
 See Sen. Dianne Feinstein, Yes: The Weapons Are Harder to Get and Police Fatalities Are Down, Insight, Feb. 26, 1996, at 26. "In 1993 . . . assault weapons . . . accounted for 8.2% of all ATF gun traces. The ban became effective on Sept. 13, 1994; from that date through November 1995, assault weapons composed only 4.3% of all gun traces." Id. See Jeffrey A. Roth & Christopher S. Koper, Wash. Urban Inst., Impact Evaluation of the Public Safety and Recreational Firearms Use Protection Act of 1994 (1997) (In contrast to most other studies discussed in this article, the Roth study relies only partially on BATF traces and includes numerous caveats about their limitations).
 See Douglas S. Weil, Traffic Stop: How the Brady Act Disrupts Interstate Gun Trafficking, (Center to Prevent Handgun Violence (1997)). http://www.handguncontrol.org/helping/trafficstopf.pdf >
 See Douglas S. Weil & Rebecca C. Knox, Effects of Limiting Handgun Purchases on Interstate Transfer of Firearms, 275 JAMA 1759 (1996).
 See Garen J. Wintemute, The Relationship between Firearm Design and Firearm Violence: Handguns in the 1990's, 275 JAMA 1749 (1996).
 See More Gun Dealers than Gas Stations, (1992).
. Fox Butterfield, Report Links Crimes to States With Weak Gun Controls, N.Y. Times, Apr. 9, 1997, at A14 (discussing Rep. Charles Schumer, War Between the States: How Gunrunners Smuggle Weapons Across America (visited Feb. 25, 1999) http://www.pcvp.org/pcvp/firearms/pubs/gunrun3.shtml>).
. It is tautological that the trace data would reflect trends in trace data. But as BATF notes, it is not appropriate to assume that trace data reflect trends for the much larger universe of untraced guns.
. See Keith Bea, "Assault Weapons": Military-Style Semiautomatic Firearms: Facts and Issues, Cong. Research Serv., Rep. No. 92-434, at 65 (1992).
. See 18 U.S.C. § 923(i) (1968).
. See 18 U.S.C. § 923(g) (1968).
. See Marianne W. Zawitz, U.S. Dep't. of Justice, Guns Used in Crime (1995) (estimating that guns were used in 1.3 million violent crimes in 1993).
. See G.L. Pierce, et al., BATF, The Identification of Patterns in Firearms Trafficking: Implications for Focused Enforcement Strategies (1996).
. See Gregore J. Sambor, Tracing Firearms, Police Chief, Mar. 1985, at 73-76.
. See id. at 73.
. See, e.g., Md. Code Ann. 27 § 36 H-3 442 (L)(1957), N.Y. Penal Code § 400 (1965); Cal. Penal Code §§ 12071, 12076-77 (1973); Mass. Ann. Laws ch. 140, § 129 C (1968).
. As explained by then Philadelphia Police Commissioner Gregore J. Sambor, "when a local agency has adequate information and their own means available, they can sometimes produce their own results quicker and with less chance of error." Sambor, supra note 13 at 73.
. See Bea, supra note 8, at 65.
. See David Freedman et al., Statistics 302-04 (1978). One of the most famous fiascos which used a non-random sample was the Literary Digest poll for the 1936 presidential election. The Digest contained the largest sample group for any poll, at 2,400,000 voters. (The Digest sent mail ballots to 10,000,000 people, using sources such as telephone books and club membership lists.) The poll predicted that Landon would defeat Roosevelt and win fifty-seven percent of the vote. However, Landon received only thirty-eight percent of the actual vote. Contrast the huge, but non-random, Literary Digest sample with the random sample of 3,000 adults typically used for the final polls taken by national polling organizations a few days before the Presidential election. These smaller, random samples almost always come within a few points of the final election result. (The major exception to this was the 1948 Truman-Dewey election, which the polls called wrong because of a misapplication of quota sampling, a topic beyond the scope of this article.)
. Bea, supra note 8, at 65.
. See D.M. Kennedy et al., Youth Gun Violence in Boston: Gun Markets, Serious Youth Offenders, and a Use Reduction Strategy, 59 Law & Contemp. Probs. 147 (1996).
. See Tom Diaz, Making a Killing 99 (1999).
. See id. at 125.
. See Kennedy, supra note 20, tbl. 5 (guns traced from youth offenders in Boston); Gerald A. Nunziato, BATF National Tracing Center, Briefing to the Homicide Research Working Group (June 10, 1997) (60 % failure rate); J. Wachtel, Sources of Crime Guns (Mar. 1996) ( paper presented at the annual meeting of the Academy of Criminal Justice Sciences in Las Vegas) (Los Angeles traces, using BATF data, supplemented by California Department of Justice data).
. Freedman, supra note 18, at 304.
. See Paul Blackman, The Uses and Limitations of BATF Tracing Data for Law Enforcement, Policymaking, and Criminological Research, 10 J. Firearms & Pub. Pol'y. 27, 62 n.19 (1998); Pierce, supra note 12, tbl. 3; Kennedy, supra note 20, at 196.
. See id.
. See CBS Morning News: Lawmakers Try to Put a Plug in Illegal Gun Pipeline (CBS television broadcast, Feb. 09, 1993).
. See Memorandum from the BATF on Firearm Statistical Information For Calendar Year 1992 to Special Agent in Charge of the New York Field Division (Oct. 22, 1992). (hereinafter Firearms Statistical Information) (on file with the author). One reason that traces from Virginia were so high was that BATF was conducting a special investigation into firearms trafficking from Virginia to New York. Thus, when a Virginia gun store would call BATF to report a suspicious sale possibly headed for New York, BATF would advise the store to go ahead with the sale. Later, when the gun-runners were arrested in New York, the gun would be traced back to Virginia. See BATF Agent Irvin W. Moran, Affidavit before U.S. Magistrate David G. Lowe (Aug. 25, 1992); Letter from John W. Magaw, Director, BATF, to Sen. Olympia J. Snowe (Feb. 23, 1996). In ordinary circumstances, the guns never would have been sold; if they had been sold, they would not have been specially selected for tracing.
. See Firearm Statistical Information. (on file with the author).
. See id.
. See S. Brill, Firearms Abuse: A Research and Policy Report (Police Found., 1977) (a handgun prohibitionist's analysis of a 1976 study of BATF traces).
. See id.
. Author's personal interview with retired N.Y. City police officer Stephen D'Andrilli (Dec. 1992) (expert in firearms licensing).
. The only non-governmental persons who may own handguns are those whose handguns were registered before 1976. See D.C. Code Ann. §§ 6-2312 (1976).
. See James D. Wright & Peter Rossi, Armed and Considered Dangerous: A Survey of Felons and Their Firearms 10-11 (1986).
. A trace to a particular state could also result from the gun's having been stolen from a firearms store, or firearms wholesaler in that state, with the gun never having been sold at retail.
. See Butterfield, supra note 6, at A14.
. Firearms Statistical Information, supra note 29.
. Bureau of Alcohol, Tobacco and Firearms, Concentrated Urban Enforcement: An Analysis of the Initial Year of Operation Cue in the Cities of Washington D.C., Boston, Mass., Chicago, Ill. 61 (1977).
. See id.
. See id.
. See 18 U. S.C. § 924 (1968).
. Virginia's law was enacted in 1994, so it is possible that a significant number of guns traced were sold before the law was enacted. South Carolina's law was enacted in the 1970s; the state's status in 1996, at least in the Schumer study, as a leading crime gun exporter, casts some doubt on the law's effectiveness. Late in the 1996 legislative session, Maryland enacted a one-handgun-per-month law, which became effective in the fall. The law could not fairly be expected to have a major impact on 1996 traces.
. Handgun Control, Inc., The Assault Weapons Ban: Questions & Answers (rev. ed. 1996). For example, one HCI brochure used a line that also appeared in print advertisements, claiming "assault weapons have quickly become the 'weapons of choice' for drug traffickers, street gangs and paramilitary extremist groups." . President Clinton repeated this line verbatim in a speech. See President William J. Clinton, Speech at the Ohio Peace Officers Training Academy ( Feb. 15, 1994).
. See David B. Kopel, Rational Basis Analysis of "Assault Weapon" Prohibition, 20 J. Contemp. L. 381, 411 (1994)(citing Jim Stewart & Andrew Alexander, Assault Guns Muscling in on Front Lines of Crime, Atlanta J.-Atlanta Const., May 21, 1989, at A1).
. See Kopel, supra note 45, at 412.
. See, e.g., Hon. Luis V. Gutierrez, Introduction of Legislation to Ban Semiautomatic Assault Weapons, H.R. 893, Cong. Rec. at E358 (Feb. 17, 1993) (citing the Cox article).
. See Kopel, supra note 45, at 412.
. See id.
. See id.
. As noted above, not all guns seized by the police are seized from criminals or are connected with a crime. For the sake of argument, I presume that every "assault weapon" seized by the police was a crime gun.
. See Gary Kleck, Point Blank 75 (1991).
. See Kopel, supra note 45, at 413.
. See id.
. See id.
. Bea, supra note 8, at 66.
. Id. at 68.
. See The Law Enforcement Technology Gun Control Survey, L. Enforcement Tech. (July/August 1991, at 14-15). Law Enforcement Technology magazine polled its readership and found that "75% do not favor gun control legislation . . . with street officers opposing it by as much as 85%." Id. In particular, 78.7% opposed a ban on "assault weapons." Id. About 37% of top management supported a ban as did about 11% of street officers. See id. Two thousand police officers participated in the magazine survey. The poll was not a random sample, but instead was based on voluntary mail responses. Accordingly, all the caveats from note 18, supra, are applicable.
. See Gun Control Act of 1968, Pub. L. No. 90-618, 82 Stat. 1313 (1968).
. See id.
. See Kopel, supra note 45. The figures were: Akron, 1 to 2%; Baltimore, less than 0.3%; Bexar County (San Antonio), 0.1%; Connecticut, 2%; Denver 0.8%; Massachusetts, 0.7%; Miami, 3.13%; Minneapolis 0.4%; Oakland, 3.9%; Orange County (Orlando), Florida, 1.9%; San Diego, 0.3%; Virginia, 3.3%. All data are from police of the relevant jurisdiction. Full citations can be found in David B. Kopel, Guns: Who Should Have Them? 179-81 (1995).
. The law sunsets in 2004. 18 U.S.C. § 922(v) & (w) (1968).
. Letter from J. Harper Wilson, Chief, Uniform Crime Reporting Section, to Paul H. Blackman (Sept. 5, 1990).
. See Michael K. Beard, For a Safer America: Ban Junk Guns, L.A. Times, Sept. 23, 1994 ("In 1995, the 3 guns the Bureau of Alcohol, Tobacco and Firearms most frequently traced to crime were junk guns") Id.; see also Coalition to stop Gun Violence, Junk Guns (August 31,1999) http://www.gunfree.org/csgv/bsc_jun.htm .>
. "(W)hen the frequency of traces is considered in proportion to each company's production, a tiny Atlanta company, S. W. Daniel, Inc., shows a tracing rate far higher." Erik Larson, The Story of a Gun, The Atlantic Monthly, Jan. 1993, at 48 (Cobray M-l1/9 pistol).
. See Weil, supra note 2.
. See Roth & Koper, supra note 1; See Handgun Control, Inc., Assault Weapons in America (visited Aug. 31, 1999) http://www.handguncontrol.org/asw.htm>; Physicians for Social Responsibility, Assault Weapons (visited Oct. 28, 1999) http://www.psr.org/assault.htm> "In the first eight months of 1995, after the ban took effect, there was an 18% drop in the number of crimes traced to assault weapons". Id.
. See Jeff Brazil & Steve Berry, Crackdown on Assault Weapons Has Missed Mark, L.A. Times, Aug. 24, 1997, at A1.
. See More Gun Dealers Than Gas Stations, supra note 5.
. See Weil, supra, note 3.
. Jeremy Bentham, Anarchial Fallacies, 2 Works of Jeremy Bentham 501 (1843).