Nancy Irene Martz appeals her conviction and sentence for distributing LSD. Martz alleges the district court1 erred in refusing to allow her to admit a California court document into evidence to impeach a key government witness. Martz also contests the district court's sentence, claiming the computation of the amount of LSD involved was erroneous. We affirm.

Postal inspectors executed a search warrant on June 26, 1990, and opened a first-class letter addressed to Paul Richard Smith in Charles City, Iowa. The letter, mailed from Oakland, California, contained 500 dosage units of LSD on blotter paper. Smith was arrested and agreed to cooperate in the ongoing investigation. Smith, acting with federal authorities in Iowa, twice wrote to Martz in Oakland requesting to purchase LSD. On both occasions, Smith received the requested LSD blotter sheets in return.

Martz was arrested and charged with three counts of distributing LSD, three counts of using the United States mails to distribute LSD, and one count of conspiracy to distribute LSD. A jury convicted Martz on all counts. The district court attributed 187.9 grams of LSD to Martz for an offense level of 36. The court found that Martz was the manager of a criminal enterprise involving more than five persons and increased Martz' offense level by three to 39. The judge also denied a two-level reduction for acceptance of responsibility. This put the total offense level at 39. With a criminal history in category I, Martz had a sentencing range of 262 to 327 months. The district court sentenced her to 288 months in prison and five years of supervised release.

During Smith's testimony, Martz' attorney cross-examined Smith about the plea agreement Smith had reached with federal prosecutors. Martz also sought to introduce evidence of two prior guilty pleas Smith had entered in California and Utah.2 Martz contended the documents would show Smith's knowledge of how cooperating with authorities could aid Smith in his own criminal case.

The district court allowed questioning about the prior pleas to the extent they demonstrated Smith's knowledge of the benefits of plea agreements and his concomitant incentive to aid prosecutors. Smith admitted in testimony that he had been charged with drug crimes in California, but he denied that he received a reduction in charges. Smith testified outside the jury's presence that he never entered a plea agreement in California, but merely pleaded guilty to two misdemeanors. The district court sustained the government's objection to the introduction of the California plea document. The court found that since the California plea required no cooperation or testimony from Smith, it gave Smith no incentive to cooperate with prosecutors and had no bearing on Smith's potential bias or prejudice. Therefore, the California document was excluded under Rule 608(b) of the Federal Rules of Evidence, which precludes the use of extrinsic evidence to prove specific instances of conduct to attack the witness' credibility. On appeal, Martz asserts the district court erred in refusing to allow introduction of the California document to impeach Smith.

Rule 608(b) gives the court discretion to allow questioning during cross-examination on specific bad acts not resulting in the conviction for a felony if those acts concern the witness' credibility. United States v. Hastings, 577 F.2d 38, 40-41 (8th Cir.1978). The rule, however, forbids the use of extrinsic evidence to prove that the specific bad acts occurred. Fed.R.Evid. 608(b). The purpose of barring extrinsic evidence is to avoid holding mini-trials on peripherally related or irrelevant matters. Carter v. Hewitt, 617 F.2d 961, 971 (3d Cir.1980) (citing 3A Wigmore on Evidence, § 979 at 826-27 (Chadbourn rev. ed. 1970)). The introduction of extrinsic evidence to attack credibility, to the extent it is ever admissible, is subject to the discretion of the trial judge. United States v. Capozzi, 883 F.2d 608, 615 (8th Cir.1989), cert. denied, 495 U.S. 918, 110 S. Ct. 1947, 109 L. Ed. 2d 310 (1990).

The district court allowed Martz to cross-examine Smith about prior guilty pleas he had made and whether he had come to realize the benefits of cutting deals with prosecutors in the past. But in conducting this questioning, Martz was required to "take his answer." Capozzi, 883 F.2d at 615; McCormick on Evidence § 42 at 92 (3d ed. 1984). While documents may be admissible on cross-examination to prove a material fact, United States v. Opager, 589 F.2d 799, 801-02 (5th Cir.1979), or bias, United States v. James, 609 F.2d 36, 46 (2d Cir.1979), cert. denied, 445 U.S. 905, 100 S. Ct. 1082, 63 L. Ed. 2d 321 (1980), they are not admissible under Rule 608(b) merely to show a witness' general character for truthfulness or untruthfulness. United States v. Whitehead, 618 F.2d 523, 529 (4th Cir.1980); James, 609 F.2d at 46. The credibility determination pertinent to the Martz trial concerned whether Smith would lie in his testimony against Martz to receive favorable treatment from prosecutors. The issue was not whether Smith, in fact, received a reduced sentence in California for pleading guilty to two misdemeanors, or whether the charges were merely dropped by prosecutors on account of lack of evidence, crowded court dockets, or other unrelated reasons. Martz' attorney argued to the district court that "a sufficient record has been made at least to establish a question for the jury at least as to whether or not a plea bargain was entered into and whether or not the defendant received the benefit of the bargain." Tr. at 192. This argument represents exactly the type of mini-trial over a collateral matter that Rule 608(b) forbids.

Martz relies on Carter, 617 F.2d 961, for the proposition that documents admitted as evidence during cross-examination of the witness do not violate Rule 608(b). Carter's holding was much narrower. In Carter,3 the Third Circuit admitted the letter in question only after the witness admitted its authenticity. The court specifically held that extrinsic evidence could not be admitted after a witness denied a charge.

[I]f refutation of the witness's denial were permitted through extrinsic evidence, these collateral matters would assume a prominence at trial out of proportion to their significance. In such cases, then, extrinsic evidence may not be used to refute the denial, even if this evidence might be obtained from the very witness sought to be impeached.

Carter, 617 F.2d at 970. Therefore, the district court did not abuse its discretion in refusing to admit the California plea document into evidence.

Martz contests her sentence based on the district court's computation of the total weight of the LSD involved. Martz contends the district court should have compiled the total weight by using the Typical Weight Per Unit Table contained in application note 11 of U.S.S.G. § 2D1.1. Utilizing this table, Martz argues, would have resulted in an offense level of 28 rather than 36.

The district court attributed 33,800 dosage units of LSD to Martz and that figure is not contested on this appeal. In computing the total weight, the district court correctly included the weight of the drug-laced blotter paper. Chapman v. United States, --- U.S. ----, 111 S. Ct. 1919, 1922, 114 L. Ed. 2d 524 (1991); United States v. Bishop, 894 F.2d 981, 985 (8th Cir.), cert. denied, --- U.S. ----, 111 S. Ct. 106, 112 L. Ed. 2d 77 (1990). The court, however, noted that blotters that were tested contained varying weights, ranging from .00692 grams per dose to .0055 grams per dose. The actual weight of only 1800 of the dosage units was known. Applying the rule of lenity, the district court attributed the lightest known weight to all dosage units and arrived at a total of 185.9 grams (33,800 doses times .0055 grams). The court added to that figure two liquid grams of LSD that were not applied to blotter paper but were attributed to Martz.4 The resulting total was 187.9 grams.

Martz argues that the district court should have applied the weight listed in the Typical Weight Per Unit Table contained in application note 11 of U.S.S.G. § 2D1.1. This table reveals a per-unit weight for LSD of .05 milligrams and would result in a total weight of 1.69 grams for the 33,800 doses. Adding in the two grams of liquid LSD and the 11 grams of LSD listed in the indictment would total 14.69 grams of LSD. This computation would have given Martz a base offense level of 28.

The district court's determination that extrapolating the lightest-known unit across the dosage units is a more reliable estimate than using the Typical Weight Per Unit Table was not erroneous. Application note 11 to U.S.S.G. § 2D1.1, itself, notes its inaccuracy and cautions that it should only be used when a more reliable estimate of weight is unavailable.

If the number of doses, pills, or capsules but not the weight of the controlled substance is known, multiply the number of doses, pills, or capsules by the typical weight per dose in the table below to estimate the total weight of the controlled substance.... Do not use this table if any more reliable estimate of the total weight is available from case-specific information.

The note provides further that the table does not include the weight of the carrying mechanism.

For controlled substances marked with an asterisk [including LSD], the weight per unit shown is the weight of the actual controlled substance, and not generally the weight of the mixture or substance containing the controlled substance. Therefore, use of this table provides a very conservative estimate of the total weight.

U.S.S.G. § 2D1.1 & comment. (n.11). Since all of these doses were on blotter paper, the weight of the blotter paper and the LSD obviously provides a more reliable estimate than the naked drug itself.

In Bishop, 894 F.2d at 987, we upheld the estimate of a total amount of LSD based on the district court's extrapolating the lightest known weight over the total number of dosage units, including those that were unrecovered. Martz attempts to distinguish Bishop by arguing that the sample of blotter paper tested in her case did not constitute a representative sample. Unlike Bishop, the blotter paper in this case did not come from the same source at the same time. Nevertheless, the district court found that there was adequate case-specific information to estimate the total weight by extrapolating the lightest known weight over all the doses.

Random testing of drugs may be sufficient for sentencing purposes. United States v. Johnson, 944 F.2d 396, 404-05 (8th Cir.), cert. denied, --- U.S. ----, 112 S. Ct. 646, 116 L. Ed. 2d 663 (1991). In Johnson, this court refused to adopt the requirement that a representative sample of drugs from each independent source be tested. See also United States v. Follett, 905 F.2d 195, 196-97 (8th Cir.1990) (estimate of drug weight permissible in plea agreement although no LSD blotters were recovered and weighed), cert. denied, --- U.S. ----, 111 S. Ct. 2796, 115 L. Ed. 2d 970 (1991).

While there may arise situations where a sample is too small or too arbitrary to extrapolate fairly over a large number of dosage units that come from disparate sources, this is not such a case. First, all of the dosage units came from Martz. Martz' bare assertion that some of the blotter sheets may have been prepared by someone else is not enough to discredit the finding that the dosage units all were distributed by Martz, consisted of LSD-laced blotter paper, and were similar in appearance. Second, in order to reduce her offense level even one step to 34, Martz would have to show that the average weight of the dosage units weighed about half of the lightest known dosage unit (.0029 compared to .0055). See U.S.S.G. § 2D1.1(c). The evidence does not show that such a wide variance is possible since the known weights were clustered at .0055 to .00692. Moreover, a cursory review of LSD blotter weights from other cases reveals that .0055 rests at the bottom of the logical range. Compare United States v. Marshall, 908 F.2d 1312, 1316 (7th Cir.1990) (en banc) (per-dose weights of .0057 grams and .00964 grams), aff'd sub nom. Chapman v. United States, --- U.S. ----, 111 S. Ct. 1919, 114 L. Ed. 2d 524 (1991); United States v. Bishop, 704 F. Supp. 910 (N.D.Iowa 1989) (per-dose weight of .0075 grams), aff'd, 894 F.2d 981 (8th Cir.), cert. denied, --- U.S. ----, 111 S. Ct. 106, 112 L. Ed. 2d 77 (1990); United States v. Andress, 943 F.2d 622 (6th Cir.1991) (per-dose weight of .0065 grams), cert. denied, --- U.S. ----, 112 S. Ct. 1192, 117 L. Ed. 2d 433 (1992); United States v. Leazenby, 937 F.2d 496 (10th Cir.1991) (per-dose weight of .0060 grams); United States v. Larsen, 904 F.2d 562 (10th Cir.1990) (per-dose weight of .0061 grams), cert. denied, --- U.S. ----, 111 S. Ct. 2800, 115 L. Ed. 2d 973 (1991); United States v. Elrod, 898 F.2d 60 (6th Cir.) (per-dose weight of .0055 grams), cert. denied, --- U.S. ----, 111 S. Ct. 104, 112 L. Ed. 2d 74 (1990); United States v. Rose, 881 F.2d 386 (7th Cir.1989) (per-dose weight of .0154 grams); United States v. DiMeo, 753 F. Supp. 23 (D.Me.1990) (per-dose weight of .0069 grams), aff'd without opinion, 946 F.2d 880 (1st Cir.1991). Therefore, we find that the district court did not err in determining that extrapolating the lightest known weight over all the dosage units was a more reliable estimate than using the bare drug weight found in the table.

We find that the district court did not abuse its discretion in refusing to admit extrinsic evidence to impugn a witness' credibility. Further, we find that the district court properly calculated Martz' sentence. The decision below, therefore, is affirmed.

HEANEY, Senior Circuit Judge, dissenting.

In my view, Nancy Martz should have been permitted to introduce into evidence two documents which established that the government informant was lying when he testified that he had not entered into plea agreements in state courts in California and Utah. With respect to drug related offenses in those states, the exhibits were not offered to prove that Smith had prior drug convictions, but rather to attack his credibility. Smith's credibility was crucial--his testimony was essential to Martz's conviction. The admission of these documents could have been accomplished quickly, and it would not have given rise to a mini-trial.

Although the Carter case well supports Martz's position, the majority distinguishes Carter on the grounds that the document in that case was admitted only after the witness admitted its authenticity. Here, however, the trial court did not ever question Smith as to the authenticity of the plea agreement. If faced with questioning about the previous plea agreements, Smith may well have backed off his previous statements, and his credibility would have been damaged.

I also believe that the majority errs in affirming the sentence. This court, over my dissent, recently held en banc that we must follow policy statements and commentary to bring about consistency in sentencing. United States v. Kelley, 956 F.2d 748, 756 (8th Cir.1992) (en banc). One would think that we would be bound by that decision where the policy statement or commentary requires a shorter sentence as well as where it requires a longer sentence.

But, apparently this is not to be the case even though the application note here is clear and precise: "If the number of doses ... but not the weight of the controlled substance is known, multiply the number of doses ... by the typical weight per dose in the table below to estimate the total weight of the controlled substance." U.S.S.G. § 2D1.1 (Application Note 11). The weight of each dose was not known; thus, the table had to be used.

Unlike the majority, I do not believe extrapolation would be proper in this case. Unlike the situation in Bishop, the blotter paper here did not come from the same source at the same time. United States v. Bishop, 894 F.2d 981, 987 (8th Cir.1990). Moreover, the amount of blotter paper weighed was a small fraction (approximately five percent) of the total amount attributed to Martz. Under these circumstances, the district court did not have enough "case-specific information" from which to make a "more reliable estimate of the total weight." U.S.S.G. § 2D1.1 (Application Note 11). Compare United States v. Shabazz, 933 F.2d 1029, 1034 (D.C.Cir.1991) (use of table in Note 11 not required where defendant conceded estimated weight of dilaudid pills was accurate, and where estimated weight was supported by data from Physicians Desk Reference, the manufacturer, and the DEA).

The majority opinion buttresses the district court's findings by favorably comparing the district court's calculation of the average weight per dose of the dosage unit (.0055 grams) to LSD blotter weights set forth in reported cases from other circuits. See ante at 791. Although the majority's review is interesting, I do not see how findings of fact from other cases can constitute "case-specific" evidence to support the district court's findings of fact in this case.

The majority also reports that a wide variance in blotter paper weights would not be possible in this case "because the known weights were clustered at .0055 to .00692." See ante at 791. With all due respect, I think this reasoning is circular: because only three samples were taken, there is no way to know whether there was a wide variance between blotter paper weights, yet the limited sample is used as proof that there was not a wide variance in weights. Moreover, there was a wide variance between even the three samples--the heaviest sample was almost twenty-five percent heavier than the lightest sample.

While it would have taken a short time to accurately determine the weight per dose, the government did not make this effort. Thus, the court was obligated to follow the table.