Defendant-appellant Robert A. Bloomer appeals from a judgment entered after a jury trial in the United States District Court for the District of Vermont (Franklin S. Billings, Jr., Judge), convicting him of one count of conspiracy to manufacture, distribute, and possess with the intent to distribute methamphetamine in violation of 21 U.S.C. Secs. 846, 841(b)(1)(A), one count of manufacturing methamphetamine in violation of 21 U.S.C. Secs. 841(a)(1), 841(b)(1)(A), one count of maintaining a place for the purpose of manufacturing, distributing, and using methamphetamine in violation of 21 U.S.C. Sec. 856(a)(1), and three counts of distributing methamphetamine to three individuals in violation of 21 U.S.C. Secs. 841(a)(1), 841(b)(1)(A). Bloomer was sentenced to 121 months' imprisonment, followed by a five-year term of supervised release.

On appeal, Bloomer challenges both his conviction and his sentencing. Regarding his conviction, Bloomer claims that the district court erred in several of its pretrial and evidentiary rulings. Bloomer also challenges the district court's denial of his motion for a new trial based on newly discovered evidence. Regarding his sentencing, Bloomer contends that the court erred in calculating the drug amount involved in his offense and in enhancing his sentence for abuse of a "special skill" pursuant to U.S.S.G. Sec. 3B1.3. The government cross-appeals, challenging certain of the district court's calculations under the sentencing guidelines.

For the reasons set forth below, we affirm the conviction, but vacate and remand for resentencing.

Beginning in 1983, defendant-appellant Robert A. Bloomer, Jr. manufactured methamphetamine, an illegal narcotic, in the chemistry lab of his Vermont home. From 1983 to 1990, Bloomer sold the drug to several customers for different periods of time. Bloomer was subsequently indicted on August 13, 1991 for six counts alleging various violations relating to his manufacture of methamphetamine. Count One alleged that Bloomer was part of a single conspiracy to manufacture and to distribute methamphetamine from 1983 to 1990. The remaining five counts alleged that Bloomer manufactured methamphetamine, maintained a laboratory for its production, and distributed the narcotics to three individuals.

Bloomer was tried by a jury before the United States District Court for the District of Vermont (Billings, J.). The evidence presented at trial included the testimony of several buyers and one distributor, and testimony by several experts on the amount of methamphetamine that Bloomer's lab was capable of producing. The expert's calculations were based on the discovery of precursor chemicals and traces of methamphetamine in the laboratory in Bloomer's home. After a two week trial, the jury found Bloomer guilty on all six counts.

Several months after the trial, Bloomer discovered new evidence regarding Dr. Brendan McMahon, one of the government's expert witnesses. Dr. McMahon, a Vermont State Police chemist, had testified that he found traces of methamphetamine on the fume hood taken from Bloomer's laboratory. The new evidence revealed that Dr. McMahon had been involved in controlled substance violations that had been unknown at the time of trial. Bloomer moved for a new trial contending that this information would have severely impeached McMahon's credibility. The district court subsequently denied this motion.

A sentencing hearing was held on December 9 and 10, 1992. Because no drugs were found in the search of Bloomer's home, the district court, for purposes of calculations under the guidelines, approximated the quantity of the drugs involved in Bloomer's offense. See U.S.S.G. Sec. 2D1.1, comment. (n. 12). Evidence was presented relating to the quantity of methamphetamine that Bloomer distributed, and the quantity that he could have manufactured at his home laboratory. The court first determined that the amount of methamphetamine that Bloomer distributed was at least 208 grams but no more than 426 grams. Next, after hearing testimony from various experts in the field of chemistry, the district court determined that in light of the precursor chemicals that were found, 504 grams of unpure methamphetamine could have been produced. Adding the two amounts together, the court found that the total amount of unpure methamphetamine that was distributed and could have been produced would be in the range of 700 to 1,000 grams. Consequently, the district court calculated the defendant's base offense level at 30. The court then determined that Bloomer possessed a special skill as a chemist and added two points pursuant to U.S.S.G. Sec. 3B1.3, resulting in a total offense level of 32. After considering Bloomer's Criminal History Category of I, the district court sentenced Bloomer to 121 months, followed by five years supervised release.

Bloomer now appeals, challenging the district court's denial of certain of his pretrial, trial, and post-trial motions. Bloomer also appeals his sentence based on the district court's double-counting, and the addition to his base offense level for special skills. The government cross-appeals, asserting that the district court erred: (1) when it concluded that the experts' testimony concerning the methamphetamine involved in Bloomer's offense referred to a mixture of methamphetamine and certain impurities rather than pure methamphetamine; and (2) when it calculated the quantity of methamphetamine that the defendant produced.

On appeal, Bloomer challenges numerous rulings relating to his conviction. His claims range from attacks on the sufficiency of the evidence of his conviction, to charges of error in certain evidentiary rulings. We have carefully examined all of his these claims and find them to be meritless. We discuss only his challenge to the district court's failure to grant a new trial based on newly discovered evidence.

At trial, the government introduced the testimony of two expert chemists who supervised the investigation of Bloomer's methamphetamine laboratory. Brendan McMahon, the Vermont State Police forensic chemist, testified that he found traces of methamphetamine in the fume hood. When Bloomer mixed the precursor chemicals in the fume hood, small deposits of methamphetamine were left behind which enabled McMahon to identify the narcotic. The government's second expert, Jack Fasanello, worked as the Drug Enforcement Administration's ("DEA") clandestine laboratory coordinator for the northeastern United States. Fasanello examined the precursor chemicals seized from Bloomer's home. He testified that the presence of a tube furnace found together with pumice and thorium nitrate could only lead to the conclusion that Bloomer was manufacturing methamphetamine. Fasanello also identified notations in the defendant's handwriting that were found during the search of the defendant's study as references to procedures for the manufacture of methamphetamine. Based on McMahon's discovery of methamphetamine traces, the chemicals Bloomer ordered and stored in the laboratory, and the laboratory equipment, Fasanello concluded that Bloomer was capable of producing three to four pounds of pure methamphetamine.

On April 15, 1992, the jury found Bloomer guilty. Shortly before sentencing, Bloomer moved for a new trial on the basis of newly discovered evidence. Bloomer had learned that on November 16, 1992, McMahon was relieved of duty because of the discovery that he had taken and used regulated drugs from the state police crime laboratory. Bloomer argued that this new evidence would have impeached McMahon's credibility at trial, and could have led to an acquittal. The district court denied Bloomer's motion, finding that the "defendant has not shown that this newly discovered evidence would probably lead to his acquittal. Persuasive independent evidence ... supported [the] defendant's conviction. Moreover, there is no indication that Dr. McMahon's trial testimony was untrue."

On appeal, Bloomer argues that the district court should have granted him a new trial because the new evidence would have impugned McMahon's veracity and led to an acquittal. He contends that McMahon was the key witness and the only one who connected the presence of methamphetamine with his private laboratory. We disagree.

Federal Rule of Criminal Procedure 33 provides that "[t]he court on motion of a defendant may grant a new trial to that defendant if required in the interest of justice." In making this determination, however, "[a] district court must exercise 'great caution' in determining whether to grant a retrial on the ground of newly discovered evidence, and may grant the motion only 'in the most extraordinary circumstances.' " United States v. Imran, 964 F.2d 1313, 1318 (2d Cir.) (quoting United States v. Di Paolo, 835 F.2d 46, 49 (2d Cir.1987)), cert. denied, --- U.S. ----, 113 S. Ct. 626, 121 L. Ed. 2d 558 (1992). The motion is not favored and will only be granted when the "new evidence ... would probably lead to an acquittal." United States v. Gilbert, 668 F.2d 94, 96 (2d Cir.1981), cert. denied, 456 U.S. 946, 102 S. Ct. 2014, 72 L. Ed. 2d 469 (1982). In other words, "[a]n appellate court must weigh whether or not there is in reality a 'significant chance' that the disclosure would have induced a reasonable doubt in the minds of enough jurors to prevent a conviction." United States v. Rosner, 516 F.2d 269, 273 (2d Cir.1975), cert. denied, 427 U.S. 911, 96 S. Ct. 3198, 49 L. Ed. 2d 1203 (1976). Moreover, we will not disturb the district court's findings of fact unless clearly erroneous, and we will not overturn the district court's decision unless it abused its discretion. See United States v. Gordils, 982 F.2d 64, 72 (2d Cir.1992), cert. denied, --- U.S. ----, 113 S. Ct. 1953, 123 L. Ed. 2d 657 (1993).

We have recently stated that when the district court learns of newly discovered evidence after a conviction, it should provide relief "if the defendant makes a showing that the evidence is in fact 'new,' i.e., it could not have been discovered, exercising due diligence before or during trial, and that the evidence is so material and non-cumulative that its admission 'would probably lead to an acquittal.' " United States v. Siddiqi, 959 F.2d 1167, 1173 (2d Cir.1992) (quoting United States v. Alessi, 638 F.2d 466, 479 (2d Cir.1980)). It is clear that the evidence concerning McMahon is "new"; neither Bloomer nor the government could have discovered the evidence before or during trial. Furthermore, because the new evidence was the only evidence impeaching McMahon's credibility, it was non-cumulative. Cf. Gordils, 982 F.2d at 72 (holding that new evidence impeaching credibility of government's key witness was cumulative of other evidence of witness' criminal activity); Gilbert, 668 F.2d at 96 (holding that new evidence impeaching government's witness was cumulative of other evidence questioning the witness' character), cert. denied, 456 U.S. 946, 102 S. Ct. 2014, 72 L. Ed. 2d 469 (1982).

The evidence, however, was clearly not so material that a new trial was required. The district court correctly found that "[p]ersuasive independent evidence ... supported the defendant's conviction." There was overwhelming testimony from six buyers of Bloomer's narcotic that established Bloomer's manufacture of methamphetamine. Furthermore, Fasanello's expert opinion indicated that the presence of certain chemicals, laboratory equipment, and Bloomer's notations could lead only to the conclusion that Bloomer was manufacturing methamphetamine. Because this evidence corroborated McMahon's testimony that Bloomer's home contained traces of the narcotic, the newly discovered evidence possibly impeaching McMahon would not have materially altered the outcome of the trial. Moreover, although the evidence may have impeached McMahon's credibility, it did not serve to contradict any of McMahon's conclusions or statements of fact. As we have previously stated, " '[t]he discovery of new evidence which merely discredits a government witness and does not directly contradict the government's case ordinarily does not justify the grant of a new trial.' " United States v. Sposato, 446 F.2d 779, 781 (2d Cir.1971) (quoting United States v. Aguilar, 387 F.2d 625, 626 (2d Cir.1967)). Accordingly, we affirm the district court's denial of the new trial motion.

Bloomer challenges his sentence on two grounds. First, he argues that the district court erred in double-counting when approximating the quantity of drugs involved in the offense. Second, he contends that the district court erred in increasing his base offense level by two levels under U.S.S.G. Sec. 3B1.3 for his use of a "special skill." The government concedes that the district court erred in approximating the quantity of drugs involved, but cross-appeals contending that the district court made additional errors that underestimated the quantity of methamphetamine involved in Bloomer's offense.

Bloomer contends that the district court misapplied the sentencing guidelines in approximating the quantity of drugs involved in his offense. He argues that by adding the amount of methamphetamine found to be distributed to the amount found to be manufactured, the district court counted the same amount of drugs twice. We agree.

In calculating the amount of drugs involved, the court first found that the evidence showed that Bloomer had distributed methamphetamine in a total amount of at least 208 grams but no more than 426 grams. Next, the court estimated that the precursor chemicals found in Bloomer's lab could reasonably have produced a quantity of 504 grams. The court then added these two amounts together to arrive at a total amount in the range of 700 to 1,000 grams. In this case, however, the addition of these two figures amounted to double-counting. The 504 figure was based on the precursor chemicals found in Bloomer's home and on all the precursor chemicals Bloomer had ever ordered dating back to 1982. It was therefore an estimate of the total production capabilities of Bloomer's lab. Because the government admitted that there is no evidence that Bloomer distributed any drugs other than those that he manufactured, the approximation of the quantity of drugs that he manufactured, the 504 figure, includes the quantity that he distributed, the figure ranging from 208 to 426.

Because of the above error, we remand the case for resentencing.

Bloomer next contends that the district court erred in increasing his base offense level by two levels pursuant to U.S.S.G. Sec. 3B1.3, which provides for an increase if the defendant "used a special skill, in a manner that significantly facilitated the commission or concealment of the offense." Bloomer claims that he did not demonstrate any special skill in his manufacture of methamphetamine because he was a self-taught amateur. We disagree.

Although the sentencing court's findings of fact regarding the special skill enhancement are subject to the clearly erroneous standard, its ruling concerning what constitutes a special skill involves a legal interpretation of the guidelines and is reviewed de novo. See United States v. Rivera, 971 F.2d 876, 893-94 (2d Cir.1992) (citing United States v. Shouldberg, 895 F.2d 882, 884 (2d Cir.1990)). See also United States v. Stroud, 893 F.2d 504, 507 (2d Cir.1990). A special skill is one "usually requiring substantial education, training, or licensing." See U.S.S.G. Sec. 3B1.3, comment. (n. 2) (emphasis added). Because the comment adds the word "usually," we find no basis for limiting the increase to only those with formal educations or professional skills. See United States v. Hummer, 916 F.2d 186, 191 (4th Cir.1990) (finding that the use of the word "usually" in the note to U.S.S.G. Sec. 3B1.3 "implies that substantial training is not a mandatory prerequisite to making a special skills adjustment"), cert. denied, --- U.S. ----, 111 S. Ct. 1608, 113 L. Ed. 2d 670 (1991). Bloomer presents the unusual case where factors other than formal education, training, or licensing persuade us that he had special skills in the area of chemistry. As noted by the government, Bloomer experimented often as an amateur chemist with explosives and fireworks. He built an extremely sophisticated home chemistry laboratory. He also used his chemical acumen professionally in teaming up with chemist Mark Malros to conduct a joint project to develop a sophisticated medical testing device. Furthermore, the presentence report indicates Bloomer took courses at various colleges, including Cornell University, and his own witness testified that Bloomer "attended college to become a chemist." As accurately stated by the district court, Bloomer was a "self-educated, strongly technical chemist, with an extraordinary sophistication." Accordingly, we find that the district court properly applied the two-level enhancement for use of a special skill.

Although the government has agreed to a remand based on the district court's double-counting, it has cross-appealed, arguing that the district court made numerous other errors in calculating the drug quantity that resulted in an underestimating of the attributable amount. Specifically, the government contends that the district court used the wrong ratio of chemicals in determining the amount of drugs that could be produced; misunderstood the government's witnesses's calculations of quantity to be impure, rather than pure, methamphetamine; and relied on an erroneous figure mistakenly proposed by a defense expert witness. Although the district court may credit the testimony of either witness, see United States v. Jacobo, 934 F.2d 411, 417 (2d Cir.1991), a careful review of the complicated testimony noted in the record indicates that the district court did confuse certain statements made by the witnesses.

The manufacture of methamphetamine involves a reaction of phenylacetic acid ("P2P") and methylamine. There are various methods of conducting this reaction and various ratios of the precursor chemicals may be used. The different methods and different ratios yield different amounts and different qualities of methamphetamine. Before it could determine the amount of methamphetamine attributable to Bloomer, the district court therefore needed to choose both a method of manufacturing methamphetamine and a ratio of precursor chemicals involved.

Regarding the method of manufacturing methamphetamine, Bloomer's expert, Dr. Brown, gave testimony concerning the "Ogato" method of methamphetamine production. In discussing the appropriate ratio to use, he contended that it would be best to use either a 10-to-1 ratio or a 5-to-1 ratio of P2P to methylamine. He calculated that using the Ogato method, a 10-to-1 ratio would yield 194 grams of methamphetamine, whereas a ratio of 5-to-1 would yield 388 grams of methamphetamine.

The government's expert, Jack Fasanello, gave testimony regarding the aluminum amalgam method of manufacturing methamphetamine. Fasanello testified that this method was used in 70% to 80% of the clandestine labs that the DEA investigated on the east coast. Fasanello claimed that in his experience the most common ratio used on the "street" was 2-to-1. Using the aluminum amalgam method and a 2-to-1 ratio Fasanello opined that, conservatively, Bloomer could have produced 3 to 4 pounds of methamphetamine from the chemicals that he ordered.

After hearing Fasanello and Dr. Brown's testimony, the district court chose the aluminum amalgam method and the 10-to-1 ratio. Dr. Brown had testified that this method and this ratio would yield 504 grams of methamphetamine.

On cross-appeal the government argues that the district court erred in choosing the 10-to-1 ratio. We recognize that the district court, in choosing a ratio, could accept the testimony of either expert. See Jacobo, 934 F.2d at 417 ("the court is to ... make its own assessments as to the weight of the evidence and the credibility of the witnesses."). The district court, however, relied in part on incorrect information in deciding to accept the calculations of Bloomer's expert. The district court held that "[a] ratio between methamphetamine and P2P of 10-to-1 is preferred, not only because it makes a cleaner product, but also because it has a greater yield, and therefore is cheaper to manufacture." Both Dr. Brown's testimony and his calculations indicate that the 5-to-1 ratio yields the higher quantity of methamphetamine. Although the district court may still choose the 10-to-1 ratio, in resentencing Bloomer it should consider that Dr. Brown has testified that such ratio would actually lead to a smaller quantity of methamphetamine being produced than a 5-to-1 ratio.

Once the quantity of methamphetamine is established the district court must follow the Drug Quantity Table of guideline Sec. 2D1.1 in determining the defendant's base offense level. The table makes reference to two categories of methamphetamine: methamphetamine and methamphetamine (actual). Methamphetamine refers to the total end result of the chemical reaction which yields a mixture of methamphetamine and various impurities. Methamphetamine (actual) refers to the part of the end product that is pure methamphetamine. See U.S.S.G. Sec. 2D1.1, comment. (backg'd). On cross-appeal the government argues that the district court misinterpreted the concept of "methamphetamine (actual)" and therefore used the wrong table in calculating Bloomer's base offense level. We agree.

It appears that the district court mistakenly thought that because both chemists testified that the procedures they were explaining would lead to a reaction that contained only a certain percentage of pure methamphetamine, the numbers that they were discussing referred to the amount of the mixed end product. It wrote:

Neither ratio produces pure or actual methamphetamine; the 10 to 1 ratio produces methamphetamine that is only 80 to 90% pure at best. Therefore, the court finds that the possible methamphetamine to be produced would not be "actual" methamphetamine.

Although at times the chemists' testimony is somewhat confusing, we find that the record indicates that in making their calculations, the chemists were referring to only the pure methamphetamine and not the total weight of the methamphetamine and the impurities. For example, on cross-examination Bloomer's expert testified that using the aluminum amalgam method a 3-to-1 ratio would yield 1.3 kilograms of methamphetamine hydrochloride. He then noted that the methamphetamine would contain "10 or so percent" of other material. The following questioning then took place:

Q. But that 10 or so percent will not be 10 percent of that 1.3 kilograms. It will be 10 percent on top of that, won't it? The 1.3 kilograms of methamphetamine is the pure methamphetamine hydrochloride that is made, isn't that correct?

A. That is correct.

Q. And if there was other byproducts in it, it would be additional weight to the 1.3 kilograms. Is that correct?

A. Correct.

The testimony of the government's chemist, Fasanello, further demonstrates the district court's error. On redirect examination at the sentencing hearing he testified as follows:

Q. Secondly, your opinion with regard to 3 or 4--the production worst case scenario, 3 or 4 pounds of methamphetamine. Is that 3 or 4 pounds of pure methamphetamine?

A. Yes.

Q. Yes, that is?

A. Yes.

Q. Okay. Now, so that is not like 90 percent pure methamphetamine or 70 percent pure--

A. Whatever the decrease in the percentage would have a commensurate increase in the weight.

Q. So that if it was 70 percent pure, there would be more of the final product, there would be 4 or 5 pounds or something like that. Is that your testimony?

A. Yes.

Accordingly, at resentencing the district court should use the methamphetamine (actual) tables in calculating the quantity of methamphetamine produced, unless the court relies on new calculations that it assures itself refer to the entire end result of the reaction.

The government's final argument on cross-appeal is that the figure that the district court chose to rely on contained an error that Dr. Brown had previously admitted to in calculating the amount of one of the precursor chemicals. On cross-examination Dr. Brown was asked to determine the amount of methamphetamine that could be produced using the aluminum amalgamate method and a 3-to-1 ratio. While going through his calculations, he was asked to explain his initial computations concerning the amount of P2P that he was using to arrive at his estimate of methamphetamine. After reviewing his notes, Dr. Brown realized that he had used the wrong yield for a particular step of the reaction dealing with the amount of P2P available. At that point, Dr. Brown recalculated his estimate of methamphetamine based on his new calculations concerning P2P.

The government argues that because Dr. Brown was never asked to recalculate the 504 gram figure following the revelation of his error, the number is "simply wrong." In contrast, Bloomer claims that such error is "harmless and without impact" because "the amount of methamphetamine which could be produced from this P2P was limited by the amount of methylamine available." While we note Dr. Brown's error in calculating the amount of P2P available, we cannot say with certainty whether the error was important in the determination of the 504 figure. Although we have learned a great deal about the manufacturing of methamphetamine in the course of reviewing the record, we are judges, not chemists. Consequently, we are not in a position to determine whether Dr. Brown's earlier error affects the 504 figure. Therefore, if the district court chooses to rely on this figure it should receive the testimony to determine whether Dr. Brown's earlier miscalculation affects the 504 figure. If the court determines that it did, the amount must be recalculated.

In sum, we remand for resentencing based on the district court's error in double counting the amount of methamphetamine involved in Bloomer's offense, and its reliance on the mixture table to measure an amount of pure methamphetamine. In resentencing, the district court should be aware that one of its asserted bases for choosing the 10-to-1 ratio suggested by Bloomer's expert is incorrect. Regardless of whether the district court chooses to accept the government's figures as noted in the record or to take further testimony, the court should not add the amount of methamphetamine found to be distributed to the amount found to have been manufactured, and should rely on the methamphetamine (actual) table unless new testimony is given concerning the mixture amount.

Based on the foregoing, we affirm defendant's conviction, but vacate and remand this case for further sentencing proceedings consistent with this opinion.