HARLINGTON WOOD, Jr., Circuit Judge.

The plaintiffs in this case represent a class of borrowers. Each class member borrowed money pursuant to ninety-day unsecured loans from American National Bank (“ANB” or “the bank”) during the relevant time period. The loan agreements provided that the borrowers were to repay the funds at the prime rate plus an additional percentage amount that varied from loan to loan (e.g., prime plus one percent). These loans are referred to as “prime-plus” loans. The loan agreements defined the prime rate as “the rate of interest charged by the bank to its largest and most creditworthy commercial borrowers (“LMCB’s”) for ninety-day unsecured commercial loans.”

At different times, ANB set its prime rate and charged the plaintiffs interest accordingly. The plaintiffs now argue that pursuant to the loan agreements the prime rate was supposed to be the lowest rate actually charged to any LMCB at any given time. They have identified seventy-three loans, made during the relevant time period, in which the interest rate was lower than the stated prime rate. Consequently they argue that by declaring a prime rate that was higher than the rates actually charged on these loans the bank breached the loan agreements and vio *1435 lated the Illinois Consumer Fraud and Deceptive Business Practices Act, 815 ILCS 505/1 et seq., and the mail fraud provisions of the Racketeer Influenced and Corrupt Organizations Act (“RICO”), 18 U.S.C. § 1962(a), (c), by engaging in a scheme to defraud prime-plus borrowers.

The district court concluded that the bank’s only responsibility was to set prime at a reasonable estimate or forecast of what it expected to charge to its LMCB’s, rather than the rates actually charged at any given time. It then analyzed the seventy-three loans in question and determined that none of them evidenced a scheme to defraud or a breach of contract. Therefore, the court granted summary judgment in the bank’s favor and awarded costs to the bank. The plaintiffs appealed both the summary judgment and the award of costs. 1 We review the grant of summary judgment de novo.

A. Prime Rate

The first issue we must discuss is whether the bank is obligated to set the prime rate based upon the actual rate of interest it charges on any particular loan. The loan agreements purportedly provide a clear definition of the prime rate — the rate the bank charges to its LMCB’s. The plaintiffs argue that the language in the agreements mandates that the prime rate must be the lowest rate the bank charges to any single LMCB at any given point in time. According to their argument, an LMCB is, by definition, the one entity that receives the lowest interest rate at any time and therefore that rate must be the prime rate. The plaintiffs cite no authority to support their definition of an LMCB and no authority to support their interpretation of the contract language. 2

The plaintiffs’ interpretation suffers from several flaws. First of all we note that “[rjeasonable bankers could disagree about which borrowers are the most creditworthy.” Kleiner v. First Nat’l Bank of Atlanta, 581 F.Supp. 955, 959 (N.D.Ga.1984). Second, the bank may very well have many customers who would fit its definition of an LMCB. Those customers may not each receive the same interest rate. Therefore there would not be one interest rate charged to all LMCB’s. This is problematic because the language of the loan agreements refers to one rate being charged to many LMCB’s. This language itself belies plaintiffs’ assertion that the prime rate is a rate charged to any particular borrower.

Furthermore, as a practical matter, a bank cannot reasonably be expected to change its stated prime rate several times a day or even on a daily or weekly basis based upon small changes or differences within the large pool of customers classified as LMCB’s. If the bank gave loans to two LMCB’s at different rates one hour apart, the plaintiffs insist that one rate would be prime for that one hour period. The bank would have to make an instantaneous announcement of its new prime rate every time minor fluctuations like that occurred. Furthermore, according to the plaintiffs’ argument, when calculating the interest owing on the prime-plus loans the bank would have to determine the amount of time each rate was in effect and calculate the interest for that time period, even if it was only a matter of hours or minutes. No reasonable person would ever expect a bank to engage in such a nonsensical bookkeeping nightmare. Such an interpretation therefore cannot be consistent with the expectations of the parties.

The bank argues that the contract language permits it to set prime at an estimate of the rate they expect to charge. The *1436 bank’s suggested interpretation is consistent with our previous holdings. We have twice before examined the issue of a bank’s duty when calculating its prime rate. In Mars Steel v. Continental Illinois Nat'l Bank & Trust Co. of Chicago, 834 F.2d 677, 682 (7th Cir.1987), we held that “a prime rate is merely a bank’s forecast of what it would charge its most creditworthy corporate customers for a 90-day unsecured loan. It is not an actual transaction price.... ” We reaffirmed that holding in United States v. Stump Home Specialties Mfg., Inc., 905 F.2d 1117, 1124 (7th Cir.1990), by stating that “a prime rate is an estimated rather than an actual price.” Relying on those holdings we conclude that the loan agreements in this case required ANB to set its prime rate based on an estimate or forecast of the rate it expected to charge its LMCB’s for any given upcoming period of time, until that forecast was later adjusted. The bank in fact satisfied its obligation of setting its prime rate at an “estimate” every time they announced a new prime rate. We therefore proceed with our analysis considering the bank’s stated prime rate to be its estimate.

B.Breach of Contract Claim

Merely concluding that the prime rate is an estimate does not end our analysis of the contract claim. Allowing the bank to set its prime rate as an estimate vests them with some degree of unilateral discretion in carrying out its contractual obligation. In such a case the common law duty of good faith and fair dealing requires that the bank exercise that discretion “reasonably, not arbitrarily, capriciously, or in a manner inconsistent with the expectations of the parties.” First Nat'l Bank of Cicero v. Sylvester, 196 Ill.App.3d 902, 910-11, 144 Ill.Dec. 24, 30, 554 N.E.2d 1063, 1069 (1st Dist.1990). 3 The plaintiffs identify seventy-three loans at rates below the stated prime and argue that those loans are sufficient to raise a genuine question of fact concerning whether the bank exercised good faith in setting the prime rate. 4

C. RICO and Consumer Fraud Claims

In addition to alleging breach of contract, the plaintiffs allege that the bank violated the Illinois Consumer Fraud Act and the mail fraud provisions of RICO by engaging in a scheme to defraud prime plus borrowers. The scheme they allege is that the bank artificially inflated its stated prime rate for the purpose of exacting higher interest charges from its prime plus borrowers. Determining whether the bank artificially inflated its stated prime rate involves the same considerations as determining if the stated prime rate is a good faith estimate of the rate the bank expected to charge to its LMCB’s. Not surprisingly, therefore, the plaintiffs argue that the same seventy-three below-prime loans mentioned above raise a genuine question of fact regarding whether the bank engaged in a scheme to defraud. In addition to showing that the bank’s stated prime was artificially inflated, the plaintiffs must also show that the bank intended to engage in a scheme to defraud. Again they contend that one can infer intent to defraud from the seventy-three below prime loans. 5

D. The Seventy-Three Loans

For purposes of determining whether ANB’s prime rate was a good faith estimate and whether the stated prime rate was artificially inflated, we must first identify those below-prime loans that are relevant ninety-day commercial unsecured loans made to the bank’s LMCB’s. Of the seventy-three loans that the plaintiffs identify as being below the *1437 prime rate the vast majority are irrelevant to our analysis.

1. Secured Loans

The plaintiffs concede that secured loans are not relevant to the establishment of a prime rate. The plaintiffs’ loan agreements clearly define the prime rate in reference to ninety-day unsecured loans. Nine of the loans identified were secured loans. In many of these loans the security was identified on the promissory note. In others the security was mentioned in related documents. In one case the security was delayed but the record makes clear that the loan officers contemplated the security at the time they approved the loan. Neither a delay in providing the security, nor listing the security in related documents does anything to affect the loans’ status as secured. Consequently, we find that these nine loans are not relevant to the issues in this case. One other loan, to Civil Aeronautics, was approved as a partially secured loan and is also irrelevant to our inquiry.

2. Personal Loans

The bank argues that twenty of the loans at issue were personal loans. It has offered evidence that at least fifteen of them were indeed loans for personal purposes and the plaintiffs offer nothing to raise a genuine question of fact about those fifteen loans. The other five are not as easily dismissed. These loans, each to Fred Eisemann, were listed as being for real estate investments and each promissory note was coded as being for commercial purposes. Thus there may be a question of fact regarding the commercial or personal status of those five loans. We need not reach that issue, however, because the plaintiffs have offered nothing to suggest that Fred Eisemann was an LMCB. 6 Therefore all twenty of these loans must be excluded from any analysis of the good faith of the prime rate estimate or the existence of a scheme to defraud.

3. Participation Loans

At various times ANB participated in loans by another bank. In such cases ANB funded a percentage of the loan that the other bank offered to one of its customers. The rate chosen for such loans was regularly the prime rate of the other bank. We therefore conclude that, as a matter of law, these loans have no relevance to whether ANB exercised good faith in establishing its own prime rate or artificially inflated that rate. Seven of the original 73 loans were such participation loans.

4. Foreign Currency

In four of the targeted loans the bank lent money in Dutch Guilders and the borrower was to repay in American Dollars. These loans have no relevance to the good faith of the bank’s interest rate. The interest rate on foreign currency loans is not comparable to the rates on standard ninety-day commercial unsecured loans because the former necessarily reflects foreign exchange rate considerations and contingencies that are not at issue in standard U.S. currency loans.

5. White Sox and Fischer Loans

One of the targeted loans, to the Chicago White Sox, was not at a rate below prime. Another loan, to Fisher Printing, was for eighty-six days. Plaintiffs concede that only loans for eighty-eight to ninety-two days are relevant.

6. LIBOR Loans

One of the plaintiffs’ main arguments is that the bank offered its best customers loans at rates below prime in the form of London Interbank Offering Rate (LIBOR) loans. By doing this the bank could maintain an artificially high stated prime rate for purposes of defrauding prime plus borrowers while giving other customers loans at its “true prime rate.” We reject this argument for two reasons. First, the record is clear that the bank made only fifteen LIBOR loans at rates below prime. When *1438 viewed in conjunction with the thousands of loans the bank made during the relevant time period, fifteen loans do not evidence a scheme to defraud or any bad faith. Even considering the fifteen loans standing alone, we find that they do not evidence a scheme to channel funds fraudulently to the banks’ best customers at lower rates. Second, all but one of those loans are distinguishable from standard ninety-day commercial unsecured loans. Eleven have already been distinguished: one was secured (B.S. Investments); one was approved as secured (Civil Aeronautics); five were participation loans (one to Brunswick and four to Maremount); and four were made in foreign currency (Handy Button).

Three other loans are also irrelevant because they were made at fixed rates. The ninety-day commercial unsecured loans that are relevant to this case are the ones offered at variable or fluctuating interest rates. Fixed rate loans are not relevant to our inquiry because they involve different economic considerations than variable rate loans. The bank does not know at the inception of a fixed rate loan if the borrower will ultimately pay more or less than a borrower on a variable loan. These contingencies and speculations affect the rates a bank will offer on a fixed rate loan. Those same considerations are not involved in variable rate loans. We have previously alluded to this idea in Mars. There the plaintiffs were prime plus borrowers who alleged that the lender bank violated the loan agreements and RICO by inflating its stated prime rate. We noted that “[tjhere [was] no suggestion that in making loans ... at fixed interest rates ... [the bank] was attempting to circumvent the terms of the loan agreements.” 834 F.2d at 679.

Only one LIBOR loan is possibly relevant. That was the loan to Merrill Lynch Panama. The bank argues that this loan is irrelevant because it was made to another bank. The plaintiffs, however, have presented evidence to suggest that loans to other banks may be considered ninety-day commercial unsecured loans. Therefore we cannot exclude this loan from our analysis.

7. Errors and Processing Delays

Any loans that resulted from clerical errors are not relevant in determining whether the bank’s stated prime rate was artificially inflated or whether it reflected a good faith estimate. Similarly, below prime loans that resulted from processing delays (e.g. the loan was agreed to on one day and the stated prime rate increased between that time and the time the loan was finally processed) are also irrelevant to the issues in this case. The bank argues that twenty-six of the seventy-three below prime loans fit into this category. Four of those loans (Handy Button, Baleor, Nelson Brothers, and Henry Valve) were fixed rate loans and will not be discussed further. See section I.D.5. For another fifteen loans ANB offered affidavits and bank officers’ memoranda demonstrating that the loans resulted from processing delays or clerical errors. The plaintiffs offer nothing to raise a genuine question of fact regarding those loans. In three of the remaining loans the promissory notes themselves disclose a clerical error.

Only four loans remain. With regard to three of them the bank has offered no documentation or evidence to obviate a question of fact. In the fourth, the bank offered an affidavit of a bank officer who stated that “in all likelihood” the below prime loan resulted from human error. This is not sufficient to eliminate a question of fact regarding that loan.

8. Summary

In total, a question of fact remains as to five loans; four that were allegedly errors, and the one loan to another bank. We may assume that these five loans are standard ninety-day commercial unsecured loans to the bank’s LMCB’s. Nonetheless five below-prime loans, out of the thousands made during the relevant time period, are insufficient as a matter of law to raise a genuine question of fact regarding the good faith of the bank’s estimate or stated prime rate. Additionally these five loans fall far short of suggesting an intentional scheme to defraud. We therefore affirm the district *1439 court’s grant of summary judgment on the contract, RICO, and Consumer Fraud claims.

II.

Discovery Requests

Through inadvertence, the plaintiffs obtained a letter from ANB’s attorneys that referred to “London documents” and London loans. The plaintiffs then sought discovery of those London documents. At that time Magistrate Judge Weisberg had assumed control over managing the discovery involved in the case. The plaintiffs argued that the letter indicated that ANB was concealing relevant information. ANB objected to the plaintiffs’ discovery request. It showed that the London branch of the bank made no loans to domestic U.S. borrowers and all loans from the London Branch were made in Eurocurrency, based on the LIBOR rate, or in Sterling, based on the Sterling Base Rate which is enforced by the Bank of England.' Judge Weisberg determined that any documents from the London branch would therefore be irrelevant to this case and denied the plaintiffs’ discovery requests. The district court affirmed that ruling. We now review that decision under the abuse of discretion standard. See Rennie v. Dalton, 3 F.3d 1100, 1110 (7th Cir.1993), cert. denied, — U.S. —, 114 S.Ct. 1054, 127 L.Ed.2d 375 (1994).

Plaintiffs argue that the standard for discovery is not relevance but whether the requested discovery “is reasonably calculated to lead to the discovery of admissible evidence.” Fed.R.Civ.P. 26(b)(1). They have correctly quoted a portion of Federal Rule 26. They omitted to quote the first part of Rule 26(b)(1) which states that “[p]arties may obtain discovery regarding any matter, not privileged, which is relevant to the subject matter involved in the pending action.” Because the London branch made loans only in foreign currency, based on foreign rates, to foreign borrowers, documents and information from that branch concerning those loans are not relevant to this case, which deals with an American bank’s domestic prime rate. Therefore we conclude that the courts below did not abuse their discretion in rejecting plaintiffs’ proposed discovery.

III.

Costs

After granting summary judgment in favor of the bank, the district court awarded costs pursuant to Federal Rule of Civil Procedure 54 and 28 U.S.C. § 1920. When the bank filed its original bill of costs the plaintiffs objected to only two items. The district court rejected both objections and awarded costs in the amount of $86,-454.51. The plaintiffs appeal that award. We can only reverse a district court’s award of costs if it resulted from an abuse of discretion. Barber v. Ruth, 7 F.3d 636, 644-45 (7th Cir.1993). The plaintiffs now raise several new objections and arguments. Because they did not raise these arguments before the district court, the bank argues that they have waived them. See Libertyville Datsun Sales, Inc. v. Nissan Motor Corp., 776 F.2d 735, 737 (7th Cir.1985). In most cases arguments not raised in the district court will be considered waived on appeal. We have, however, recently recognized an exception to this rule for plain errors in civil eases. Amcast Indus. Corp. v. Detrex Corp., 2 F.3d 746, 749 (7th Cir.1993), cert. denied, — U.S. —, 114 S.Ct. 691, 126 L.Ed.2d 658 (1994). So long as the new argument is a “pure issue of statutory interpretation” and the issue is fully argued in the briefs, we can decide the issue for the first time on appeal. Id. at 749-50. We therefore examine the plaintiffs’ arguments, decide which are pure issues of statutory interpretation, and determine whether the particular awards of costs resulted from an abuse of the district court’s discretion.

A. Costs for Printing Briefs to the Seventh Circuit and United States Supreme Court

This litigation has seen many turns. Initially the plaintiffs’ complaint was dismissed. They appealed that dismissal to this court and we reversed in Haroco, Inc. v. American Nat’l Bank & Trust Co. of Chicago, 747 F.2d 384 (7th Cir.1984). The bank sought and received review of that decision in the United States Supreme Court. The *1440 Supreme Court affirmed in American Nat’l Bank & Trust Co. of Chicago v. Haroco, Inc., 473 U.S. 606, 105 S.Ct. 3291, 87 L.Ed.2d 437 (1985). Following that decision the case was remanded to the district court. When the bank ultimately prevailed on summary judgment, it asked for an award of the costs associated with preparing and filing its briefs to this court and the Supreme Court in the earlier appeals, in which incidentally it was not the prevailing party. The district court awarded the bank $4076 to print its brief before this court and $12,985.83 to print its briefs in the Supreme Court. We now conclude that both awards were erroneous as a matter of law.

Neither the rule nor the statute that the district court relied upon in granting costs mentions costs associated with filing appeals. See Fed.R.Civ.P. 54; 28 U.S.C. § 1920. This is not by accident. The procedure for recovering the costs of printing appellate briefs is clearly provided in Federal Rule of Appellate Procedure 39. That rule mandates that the party seeking such costs must file a bill of costs within fourteen days after the entry of judgment. Fed.R.App.P. 39(d). The bank filed its first requests for such costs more than nine years after the judgment. Furthermore, we note that the request for such costs is to be filed with the Court of Appeals, not before the district court. Finally, even if the bank had timely filed its request it would have been rejected outright because it was not the prevailing party on that appeal. The rule makes clear that “if a judgment is reversed costs shall be taxed against the appellee.” Fed.R.App.P. 39(a). Here the bank was the appellee. Not only was it not entitled to costs, it could have been responsible for the plaintiffs’ costs. As a matter of law the district court erred by awarding these costs and the ultimate award must be reduced by $4076.

Similarly the award of costs for printing the briefs in the Supreme Court must be reversed. Once again neither § 1920 nor Federal Rule 54 mention anything about the costs associated with appeals to the Supreme Court and again this is not by accident. Such costs are addressed by the Supreme Court rules. Those rules state that “the expenses of printing briefs ... are not taxable.” U.S.Sup.Ct.R. 43.3. Therefore even if the bank had been the prevailing party in the Supreme Court, it would not have been entitled to the costs of printing its briefs. It should not be able to achieve an end run around that rule by seeking those costs in the district court. The award of costs must be reduced by $12,985.83.

B. Costs of Computerized Legal Research

The district court awarded the bank $30,439.57 for computerized legal research as an element of costs. The plaintiffs now argue that this was improper. They did not raise their argument in the district court. Nevertheless we find this to be an issue of pure statutory construction and an important one at that. Computerized legal research involves an attorney sitting down in front of a computer and researching legal issues by searching through a database which now includes almost every resource one would find in the country’s largest law libraries. In addition to the attorney charging the client for the time he or she spends doing this research, the companies that offer the computerized legal research services also charge a fee. Theoretically, even though the clients now pay two fees, their ultimate bill should be lower because the attorney should be able to do the research more quickly and efficient ly. If this research had been done manually by an attorney sitting in the library reading through books rather than sitting before a computer screen, nobody would dispute that the attendant fees would be properly classified as attorney’s fees and not costs.

Recognizing this, we have previously held that computer research costs “are more akin to awards under attorney’s fees provisions than under costs.” McIlveen v. Stone Container Corp., 910 F.2d 1581, 1584 (7th Cir.1990). In fact such costs are indeed to be considered attorney’s fees. The added cost of computerized research is normally matched with a corresponding reduction in the amount of time an attorney must spend researching. Therefore, we see no difference between a situation where an attorney researches manually and bills only the time *1441 spent and a situation where the attorney does the research on a computer and bills for both the time and the computer fee. In both cases the total costs are attorney’s fees and may not be recovered as “costs.” Therefore the award of costs in this case must be reduced by $30,439.57.

C. Exemplification Fees

Included in the award of fees to the bank was $12,389.71 for exemplification fees for the preparation, copying, and collating of an exhibit to the bank’s motion for summary judgment and also fees for graphics services. The plaintiffs now argue that copying and collating documents is work that an attorney or paralegal should do and thus is to be considered under the attorney’s fees statutes. We reject that argument outright. Collating documents does not require the skills of a licensed member of the bar or a paralegal. Furthermore the statute speaks to this issue unequivocally. The court may tax “[f]ees for exemplification and copies of papers necessarily obtained for use in the case.” 28 U.S.C. § 1920(4). We find that the costs for the copying and collating of exhibits and the graphics charges fall squarely within this language and the award was proper in this regard. We do not address plaintiffs’ arguments that the charges were unreasonably high or unsubstantiated because they did not raise those issues before the district court and the issues are factual and wholly unrelated to statutory interpretation.

D. Copying Charges for Pleadings and Discovery Documents

The bank’s award of costs included $21,986.40 for copies of discovery documents and pleadings at a rate of twenty cents per page. This was the rate the bank’s attorneys charged for copies made at the law firm. The plaintiffs argue first that the award includes a set of copies that is non-compensa-ble because it was for the attorney’s convenience. We agree. The district court recognized that only two sets of copies would be reasonable. It therefore reduced the award by $1000. Nowhere did it explain why $1000 was a reasonable reduction for the third set of copies. We remand this issue back to the district court for a more precise determination of the number of copies that were unnecessary.

The second argument the plaintiffs raise with respect'to these copying costs is that the twenty cent per page charge is unreasonably high. The attorneys charged twenty cents for copies made in-house. We have previously held that “charges for in-house reproduction may not exceed the charges of an outside print shop.” Martin v. United States, 931 F.2d 453, 455 (7th Cir.1991). Although our holding in Martin addressed copying charges under Federal Rule of Appellate Procedure 39, we find it equally applicable to copying charges awarded in the district court. The district court did not make a finding as to the outside print shop charges. We therefore remand to the district court for a determination of a reasonable rate for copies charged by outside print shops and direct that the award of costs be adjusted appropriately.

E. Deposition Transcripts and Court Reporter Fees

A total of thirty-six depositions were taken in this ease. The plaintiffs initiated thirty-five of those thirty-six depositions. They selected the court reporters and therefore selected the court reporters’ fees. They now object to the district court awarding the bank costs for transcripts of those depositions. Their principal argument is that the bank obtained copies of transcripts rather originals; therefore it should receive a lower rate per page (i.e. the copy rate). They did not make this argument before the district court and they cited us nothing in the record that would indicate that the bank obtained copies of the transcripts. We note that the bank was awarded costs for only one set of transcripts. Furthermore the bank filed the depositions in this ease. Therefore we find that the district court did not abuse its discretion in finding that the bank obtained original transcripts and awarding costs at the original transcript rate. The plaintiffs’ other argument is that the award should reflect the rate per page that was in effect when the deposition was taken, not *1442 when it was ordered. This is unreasonable. The cost of a transcript does not exist until the deposition is transcribed. The date the deposition was taken is irrelevant. We therefore affirm the district court’s award of costs for the deposition transcripts.

F. Witness Fees

The plaintiffs raise two issues with respect to the award of costs for witness fees. First they claim that the district court was not permitted to award fees, in the amount of ninety dollars, for Ronald Grayheek’s attendance at his own deposition because he was then a party to the suit. Although they did not raise this argument below, we address it because it is one purely of statutory interpretation. The relevant statute addressing the issue of costs allows costs for witness fees. 28 U.S.C. § 1920(3). It admittedly makes no distinction between party and non-party witnesses. The general rule, however, is that “parties may not normally collect witness fees.” Barber v. Ruth, 7 F.3d 636, 646 (7th Cir.1993). It follows that the district court may not tax witness fees for party witnesses under 28 U.S.C. § 1920(3). The awarding of costs for Ronald Grayheck’s appearance at the deposition was therefore improper and the total award must be reduced by ninety dollars.

Next the plaintiffs argue that the district court abused its discretion by awarding the fees paid to witnesses who were subpoenaed and given notices of depositions but whose depositions were never taken. We have addressed a similar issue in Spanish Action Comm. of Chicago v. City of Chicago, 811 F.2d 1129 (7th Cir.1987). In that case the prevailing party sought witness fees for witnesses who were subpoenaed to testify at trial but were never called. The trial judge denied the request for those costs. We reversed holding that the district court abused its discretion by denying the costs because the witness fees compensated the witnesses for their availability and readiness to testify rather than actual testimony. Id. at 1138. Here we follow the same principle. The subpoenaed witnesses were compensated for their readiness to testify at the depositions. The award of costs for these witnesses was proper.

The plaintiffs do not object to the award of $340 in docket fees or $925.75 for transcripts of hearings. As a consequence that portion of the costs award is affirmed.

IV.

The decision of the district court granting summary judgment in favor of American National Bank is Affirmed. The award of costs is AffiRmed in part, Reversed in part and REMANDED. In this court the parties are to bear their own costs.

. The plaintiffs also appeal the district court’s denial of their request for additional discovery. We will address that issue separately below in section II.

. The plaintiffs argue that Ronald Grayheck, an officer of ANB, admitted in his deposition that the prime was the rate actually charged to LMCB's. The plaintiffs’ use of this deposition testimony is creative at best. The quotes they extract come from a long, rambling, and unintelligible series of questions. Mr. Grayheck warned that he had "lost his train [of thought].” The plaintiffs’ own attorney who was asking the questions confessed that the rambling question had "taken on a life of its own.” We decline to ascribe any meaning to responses from such a series of questions.

. This effectively answers the plaintiffs' concerns that the Mars interpretation would allow the bank to set its estimate at any rate they chose, however high or unreasonable.

. In their brief the plaintiffs argue that any single loan below prime would be a breach of contract. Given our holding in section I.A., this cannot be true. All loans below prime may be useful in analyzing whether the estimate was made in good faith, but they cannot individually be dis-positive of the issue.

.The plaintiffs insist that they need show only two loans below prime to satisfy RICO. Again we note that the issue is not whether any particular loan was below prime but whether the estimate, or stated prime rate, was artificially inflated.

. The plaintiffs maintain that any person receiving an interest rate below prime is an LMCB. As we mentioned earlier, plaintiffs offer no authority for such a definition of LMCB and we therefore reject their argument.