A lone gunman robbed the Fifth Third Bank in Indianapolis on the morning of January 28, 1991. Holding a handkerchief over part of his nose and mouth, the robber pointed a long-barreled, dull-finish revolver at one of the tellers and demanded that she place all of the money at her teller station into a bag. She did not have enough. The robber spotted the bank's office manager, directed him over to the teller line, and ordered him to continue filling the bag from another station. Eventually, the cash from an automatic teller machine in the rear of the bank was also emptied into his sack. The robber then prepared his getaway. Taking the manager out to the parking lot, the robber told him to start his car and climb into the passenger seat. They drove a short distance, and the robber ordered him out of the car. The car was later recovered a short distance away.

Seven weeks later, a similar robbery occurred at the Shelby Federal Savings Bank in Indianapolis. A gunman entered the bank and walked toward one of the tellers, placing a handkerchief over his nose and mouth when he arrived at the window. He pointed a long-barreled revolver at the teller and ordered her to get a bag and fill it with money. This time, the branch manager arrived with a trashcan liner and began stuffing cash inside. When the teller stations ran out, the robber sent her to the vault to get more. The branch manager had recently bagged fifteen hundred dollars in one-dollar bills, and she dropped these bills into his bag. The robber then approached an elderly bank customer, Homer A. Doriot, put the bag in his hands, and led him out to the parking lot. Outside, the robber took Doriot's keys, put him in the passenger seat, and drove some distance before ordering Doriot out. The car was found the same day and returned to its owner.

While driving home on the afternoon of the robbery, Homer Doriot noticed a pair of eyeglasses that he did not recognize on the front seat of his car. Doriot contacted the bank manager, who notified the FBI. The next day, Special Agent Roth of the FBI retrieved the glasses from Doriot's car. That started an impressive performance of high-tech sleuthing. Roth delivered the glasses to Edward H. Schmidt & Sons, an Indianapolis company in the business of manufacturing custom eyeglasses. Mark Schmidt, its president, had the eyeglasses electronically "neutralized," or analyzed, by a machine that reads each lens's prescription by means of refracted light. Schmidt also determined that the glasses had "Geoffrey Beene" designer frames, roughly eight months to a year old. Armed with this information, Roth went looking for a retailer that sold Geoffrey Beene eyeglasses. He contacted City Optical Company, a corporation that operates "Dr. Tavel Premium Optical" and "Dr. Tavel Vision Value" eyeglasses chains in Indianapolis. Their records turned up a total of seventeen sales of the particular type of frames found in Doriot's car, only one of which matched the prescription Schmidt & Sons analyzed. The name of its purchaser was Charles William Blackburn. Within a few days of the second robbery, almost as if he wanted to confirm the FBI's theory, Blackburn himself walked into a Tavel retail outlet, looking for a replacement pair of eyeglasses.

On the day of the second robbery, Blackburn visited Crickett Yevone Robertson, an exotic dancer, at the bar where she worked. Robertson testified that "Bill" Blackburn knew her from his regular visits to the establishment. Blackburn said that he wanted to take her shopping for her birthday, but they ended up rescheduling their date for later in the week. Three days later, Blackburn handed Crickett Robertson the keys to his car and said that he had a present for her. Inside the car, she found one-dollar bills strewn all over the front seat and floor. She gathered between fifty and one hundred bills, stuffed them in her shirt, and kept the money as a birthday present.

A jury convicted Blackburn of two counts of bank robbery and two counts of use of a firearm during the commission of a crime of violence. The evidence against Blackburn was substantial. In addition to the eyeglasses match, all three bank tellers who confronted the robber identified Blackburn as the man behind the long-barreled gun. A search of Blackburn's residence turned up a worn, western-style revolver, thirty-one one-dollar bills in the pocket of a coat, and tan overalls and a gray hooded sweatshirt similar to clothing worn by the bank robber. The government also presented Crickett Robertson's testimony about her unusual gift from Blackburn. Blackburn received a heavy sentence, as violators of the firearms statute (especially those convicted on multiple counts) usually do. The district court sentenced him to concurrent terms of 125 months' imprisonment for the bank robbery counts, and to five-year and twenty-year consecutive terms for the firearms counts. He was also ordered to pay full restitution of $29,574.00 to the two banks that he robbed.

Blackburn raises three objections to his conviction. He argues first that the district court erred by refusing to define the concept of reasonable doubt to the jury, even after the jury asked the judge for a definition during their deliberations. Blackburn concedes that we have instructed district courts not to define reasonable doubt, but he maintains that the totality of the circumstances in this case--in particular, the use of "questionable" eyewitness identification and of circumstantial evidence to prove his guilt--made a supplemental instruction necessary to avoid jury confusion. According to Blackburn, the district court's refusal to define reasonable doubt amounted not merely to an abuse of its discretion, but to a full-blown violation of his Due Process rights under the Fifth Amendment. He contends that the jury must have been unclear about the meaning of reasonable doubt, or else it would not have asked for additional explanation; and if it did not understand the standard it was to apply, then it may have unfairly convicted him.

Blackburn's argument fails. We have reiterated time and again our admonition that district courts should not attempt to define reasonable doubt. See, e.g., United States v. Bardsley, 884 F.2d 1024, 1029 (7th Cir.1989); United States v. Hall, 854 F.2d 1036, 1038-39 (7th Cir.1988); United States v. Glass, 846 F.2d 386, 387 (7th Cir.1988); United States v. Martin-Trigona, 684 F.2d 485, 493 (7th Cir.1982); United States v. Regilio, 669 F.2d 1169, 1178 (7th Cir.1981), cert. denied, 457 U.S. 1133, 102 S. Ct. 2959, 73 L. Ed. 2d 1350 (1982); United States v. Shaffner, 524 F.2d 1021, 1023 (7th Cir.1975), cert. denied, 424 U.S. 920, 96 S. Ct. 1126, 47 L. Ed. 2d 327 (1976); United States v. Lawson, 507 F.2d 433, 442 (7th Cir.1974), cert. denied, 420 U.S. 1004, 95 S. Ct. 1446, 43 L. Ed. 2d 762 (1975). The pattern jury instructions of this circuit themselves state that "[t]he phrase 'reasonable doubt' is self-explanatory and is its own best definition. Further elaboration 'tends to misleading refinements' which weaken and make imprecise the existing phrase." 1 Federal Criminal Jury Instructions of the Seventh Circuit § 2.07 (1980) (citation omitted). Nothing about Blackburn's case or the way the government proved it made it improper for the district court to decide that "reasonable doubt" must speak for itself.

We have dealt before with a situation in which a jury specifically requested a definition of reasonable doubt. In United States v. Glass, 846 F.2d 386, 387 (7th Cir.1988), the jury asked that reasonable doubt be defined, the district court refused, and we upheld its refusal largely on the grounds stated above. Blackburn points out that Judge Posner, in a concurrence to another case, questioned whether we should condemn so absolutely all attempts to define the term. See United States v. Hall, 854 F.2d 1036, 1043 (7th Cir.1988) (Posner, J., concurring). Perhaps, Judge Posner suggested, there are cases in which a carefully-worded instruction--say, one modeled after the Federal Judicial Center's Pattern Criminal Jury Instructions--could provide useful guidance to a jury in its deliberations. In those cases, he opined, a judge should be allowed to give it. While this view has some appeal, it is not the law of this circuit. We still frown on definitions of reasonable doubt, given their likelihood to confuse juries more than the simple words themselves. But even if there were situations in which a district court's decision to explain reasonable doubt was justified, the choice not to define it, as occurred here, would not be in error.

Blackburn next contends that the district court abused its discretion by allowing the government to introduce Crickett Robertson's testimony that Blackburn gave her approximately one hundred one-dollar bills. He complains that portions of her testimony--especially a statement that Blackburn frequented the bar where she worked--were unfairly prejudicial because they invited the jury to impugn his character. Blackburn insists that this evidence was irrelevant to the question of whether he committed the bank robberies with which he was charged.

These arguments are also unconvincing. We review evidentiary rulings on matters of relevance under a deferential standard. See United States v. Hoffman, 806 F.2d 703, 708 (7th Cir.1986), cert. denied, 481 U.S. 1005, 107 S. Ct. 1627, 95 L. Ed. 2d 201 (1987). Federal Rule of Evidence 403 states that "[a]lthough relevant, evidence may be excluded if its probative value is substantially outweighed by the danger of unfair prejudice...." Fed.R.Evid. 403. Here, the question and answer about Blackburn's patronage of the bar supplied background narrative material linking Blackburn and Robertson. While some prejudice may have resulted, we do not believe it "substantially outweighed" the evidence's probative value for supporting the government's theory of what happened to the money Blackburn stole.1 The district court also limited any prejudicial impact the evidence might have by sustaining objections to further questions about Blackburn's visits to the bar.

The fact that evidence casts the defendant in an unsavory light is not enough to create error by its admission, as the case of United States v. Ewings, 936 F.2d 903, 907 (7th Cir.1991), teaches. In Ewings, a priest was on trial for defrauding insurance companies by taking out policies on his niece's life, while concealing the fact that she had a terminal illness. Evidence was admitted that some of the proceeds from these policies found dubious outlets--cars, a boat, a trip to Las Vegas, and gifts to "a female acquaintance of long-standing." Id. at 905. We recognized that the evidence suggested the defendant "was not a paragon of virtue." But we believed that it was relevant to show that the defendant, rather than someone else, committed the fraud. "Spending sprees, like other evidence of pecuniary gain, tend to show participation in crimes where financial enrichment is the motive." Id. at 906. We therefore upheld its introduction. The evidence in this case was admissible for the same reason.

Lastly, Blackburn argues that the district court should not have admitted two of the government's exhibits. Exhibit 4 was the pair of eyeglasses analyzed by Schmidt & Sons, after being found in Doriot's car. Blackburn argues that the eyeglasses should never have been admitted into evidence, because Doriot did not testify that Exhibit 4 was the same pair that he found in his vehicle. This argument is without merit. Special Agent Roth testified that Exhibit 4 was the same pair of glasses that he retrieved from the front seat of the car and took to Schmidt & Sons' offices. Mark Schmidt also testified that Exhibit 4 was the same pair of glasses that the FBI had delivered to him. There is no rule that the person who first discovers a piece of evidence, rather than the person who retrieves it from its resting place on behalf of the government, must lay the evidentiary foundation for its admission. Even if this evidence raised a chain of custody problem, it would go its weight, not its admissibility. See United States v. Lott, 854 F.2d 244, 250 (7th Cir.1988).

Exhibit 11 was a pair of computer printouts of the lensometer readings for each lens of the eyeglasses, with an attached hand-written prescription summarizing information about the glasses, including physical descriptions of the frames and lenses and the results of the neutralization test. Mark Schmidt testified that he met FBI agents Rice and Roth when they visited Edward H. Schmidt & Sons seeking to have the eyeglasses analyzed. Schmidt brought the eyeglasses to Bob Shirley, the company's supervisor of quality control, whose job is to operate their Nidek Marco Automatic Lensometer machine. Shirley analyzed the glasses in the presence of Schmidt and the FBI agents. Tr. II-442. Schmidt then took the printout to Phyllis Johnson, supervisor of Schmidt & Sons' customer service department, who filled out a prescription form by hand, based on the lensometer results and other observations about the glasses. Tr. II-441.

Over Blackburn's objection, the district court admitted exhibit 11 as a record of regularly conducted activity. See Fed.R.Evid. 803(6).2 Blackburn now renews his objection that the evidence was inadmissible hearsay. He points out that Schmidt did not generate any portion of the analysis in the exhibit. More importantly, Blackburn claims that the exhibit was not "kept in the course of a regularly conducted business activity." Id. Rather, it was prepared at the request of the FBI, in anticipation of a prosecution against him. For that reason, he claims it could not be a business record.

We agree with Blackburn. Clearly, the report in this case was not kept in the course of a regularly conducted business activity, but rather was specially prepared at the behest of the FBI and with the knowledge that any information it supplied would be used in an ongoing criminal investigation. As we discuss below, the method by which these documents were generated was, to be sure, highly reliable. Indeed, if the FBI had requested a copy of a lensometer report that Schmidt & Sons innocently prepared on behalf of a customer, we would have no qualms about its admissibility under Rule 803(6). Records prepared and kept in the ordinary course of business are presumed reliable for two general sorts of reasons. See Michael H. Graham, Handbook of Federal Evidence § 803.6, at 867 (3rd ed. 1991). First, businesses depend on such records to conduct their own affairs; accordingly, the employees who generate them have a strong motive to be accurate and none to be deceitful. Second, routine and habitual patterns of creation lend reliability to business records. See United States v. Rich, 580 F.2d 929, 938 (9th Cir.), cert. denied, 439 U.S. 935, 99 S. Ct. 330, 58 L. Ed. 2d 331 (1978). But when a document is created for a particular use that lies outside the business's usual operations--especially when that use involves litigation--neither of these justifications for admission holds. In finding this report inadmissible under Rule 803(6), we adhere to the well-established rule that documents made in anticipation of litigation are inadmissible under the business records exception. See, e.g., AMPAT/Midwest, Inc. v. Illinois Tool Works, Inc., 896 F.2d 1035, 1045 (7th Cir.1990); Abdel v. United States, 670 F.2d 73, 75 (7th Cir.1982); see also Palmer v. Hoffman, 318 U.S. 109, 63 S. Ct. 477, 87 L. Ed. 645 (1943); Fed.R.Evid. 803(6) advisory committee's note ("Absence of routineness raises lack of motivation to be accurate.").

The admissibility of laboratory or scientific reports in criminal cases raises another sort of problem under the Federal Rules of Evidence. Rule 803(8) allows the admission of records setting forth matters observed pursuant to a legal duty to report and factual findings resulting from investigations made pursuant to authority granted by law, except in criminal cases against the accused.3 The apparent concern of the drafters of the rule was that use of records in criminal cases would cause "almost certain collision with confrontation rights." Fed.R.Evid. 803(8) advisory committee's note. The trouble with the formal structure of the Rule 803 exceptions is that only Rule 803(8) contains this restriction on the use of investigatory documents in criminal cases. None of the other exceptions, such as the business records exception, limit the use of documents in criminal cases. But if a document prohibited under Rule 803(8)(B) or (C) can come into evidence under Rule 803(6), then the 803(8) restrictions are rendered nugatory. The question naturally arises whether the qualifiers in Rule 803(8)(B) and (C) should be read into the other hearsay exceptions.

The leading case giving an affirmative answer is United States v. Oates, 560 F.2d 45 (2d Cir.1977). There, the government sought to introduce against a criminal defendant the official report and handwritten worksheet of the Customs Service chemist who analyzed a white powdery substance that had been seized from him. The chemist himself did not testify; in his stead appeared a different Customs chemist, who had not worked with the preparer of the worksheet personally, but had conducted hundreds of similar tests. The Second Circuit ruled that it was reversible error to admit the report and worksheet as business records, holding that even if the exhibits fell within the language of Rule 803(6), the restriction in Rule 803(8) overrode it. The court reasoned that reports made by government chemists contain "factual findings resulting from an investigation made pursuant to authority granted by law," Fed.R.Evid. 803(8)(C), and the chemists themselves qualify as "law enforcement personnel" whose scientific determinations are "matters observed pursuant to duty imposed by law," Fed.R.Evid. 803(8)(B). Id. at 67; see also United States v. Cain, 615 F.2d 380, 382 (5th Cir.1980) (following Oates ).

The Eighth Circuit, on the other hand, has apparently rejected the logic of Oates. In United States v. Baker, 855 F.2d 1353, 1359 (8th Cir.1988), cert. denied, 490 U.S. 1069, 109 S. Ct. 2072, 104 L. Ed. 2d 636 (1989), the court reaffirmed its long-standing view that "[w]hen made on a routine basis, laboratory analyses of controlled substances are admissible as business records under Federal Rule of Evidence 803(6)." Id. at 1359 (citing United States v. Scholle, 553 F.2d 1109, 1124 & n. 4 (8th Cir.), cert. denied, 434 U.S. 940, 98 S. Ct. 432, 54 L. Ed. 2d 300 (1977)). In that case, the court gave little explanation for its decision, stating simply that there was no challenge as to the reliability of the reports, nor to the fact that they were made on a routine basis. Id.

Our own views on this issue have evolved. Almost 40 years ago, in a criminal case involving heroin distribution, this circuit ruled that memoranda prepared by a government chemist summarizing the results of drug tests were admissible under the predecessor statute to the business records rule. See United States v. Ware, 247 F.2d 698, 699-700 (7th Cir.1957). We explained that the probability of their trustworthiness justified the documents' admission despite the overarching ban on hearsay. Id. In more recent cases that superseded Ware, we have steered a middle course between the Eighth Circuit's wholesale admission of laboratory reports and the Second Circuit's virtual ban. In United States v. King, 613 F.2d 670 (7th Cir.1980), a case in which a postal worker was convicted of making false statements in his application for Supplemental Security Income, we held that forms prepared by claims representatives in connection with his application were admissible under the business records exception. So long as the maker of the report is available for cross-examination, we ruled, admitting the reports themselves does not contravene Rule 803(8). Cf. United States v. Sawyer, 607 F.2d 1190, 1193 (7th Cir.1979) (approving admission of memorandum prepared by IRS agent in criminal case under Rule 803(5), where agent was available for cross-examination, according to similar analysis).

Here, we believe that the restrictions contained in Rule 803(8)(B) and (C) do not bar the admission of exhibit 11. While the Second Circuit held that government chemists are "without question, important participants in the prosecutorial effort," Oates, 560 F.2d at 68, we think a private company that conducts tests independently with its own equipment and prepares reports by itself based on those tests, even if at the instance of a governmental agency, is not part of the prosecution team. The reports did not set forth "matters observed pursuant to duty imposed by law," nor were the preparers themselves vested with legal authority to pursue the investigation. Given these factors, we think it is unreasonable to exclude the documents. Cf. 4 Jack B. Weinstein & Margaret A. Berger, Weinstein's Evidence p 803(6), at 803-207 to -208 (1992) (criticizing expansive reading of Oates rule). For these same reasons, however, exhibit 11 cannot be admissible under Rule 803(8) itself. If Schmidt & Sons was not the equivalent of government investigators, then its reports were not "[r]ecords ... of public offices or agencies." Fed.R.Evid. 803(8). In short, exhibit 11 is not a public record or report.

The basic trustworthiness of this computer-generated printout, and the accompanying handwritten transcriptions, is not contested. The test Schmidt & Sons conducted was entirely mechanical and objective. At trial, Mark Schmidt explained that there are two types of lensometers: a traditional type, resembling a microscope, which requires the operator visually to focus on and measure certain "meyer" lines; and an electronic version, used here, which automatically measures the lines itself and prints out the results. Using the electronic lensometer produced results that are extremely trustworthy. See Laird C. Kirkpatrick, Confrontation and Hearsay: Exemptions from the Constitutional Unavailability Requirement, 70 Minn.L.Rev. 665, 701 (1986) ("The higher the reliability of the testing procedure, the lower the necessity for confrontation. The greater the objectivity of the test, the less its susceptibility to testing by cross-examination."). Furthermore, there was little possibility of bias against Blackburn, since the criminal investigation at that time had no suspects. The purpose of the lensometer tests was not to build a case against any specific individual, but merely to determine the prescription of the glasses found in Doriot's car.

These considerations suggest that the exhibit was reliable enough to be admissible under the residual hearsay exception, Rule 803(24).4 The other requirements of the rule are satisfied as well. Subsection B states that the admitted evidence must be more probative on the point for which it is offered than any other evidence that the proponent reasonably can procure. On these facts, the machine-created report is at least as probative of the results of the electronic neutralization test as any oral statement that could be offered by an observer of the same test. Accordingly, the admission of the exhibit was proper.

For the foregoing reasons, Blackburn's conviction is

AFFIRMED.