On February 12, 1996, after a ten we’ek trial, a jury found the three defendants, Baldev Raj Bhutani, Neelam Bhutani, and ALRA Laboratories, Inc., guilty of a number of charges related to their manufacturing and distribution of mislabeled and/or adulterated generic pharmaceutical products. After various post-trial motions were denied, the defendants filed a motion for a new trial based on newly discovered evidence and/or the government’s violation of Brady v. Maryland, 373 U.S. 83, 83 S.Ct. 1194, 10 L.Ed.2d 215 (1963). The district court granted the defendants’ motion for a new trial on all counts of conviction based on the government’s violation of Brady. The government now appeals that ruling. For the reasons set forth below, we reverse.
I. BACKGROUND
Baldev and Neelam Bhutani are the husband and wife team that owns ALRA Laboratories, Inc., a pharmaceutical company that produces generic prescription drugs. The trial primarily focused on the manufacturing and distribution process of two drugs produced by ALRA — Lactulose and K+10. Lactulose is a drug used to combat advanced liver disease, and K+10 is a potassium supplement.
On January 26, 1994, a federal grand jury filed a superseding indictment charging all three defendants with violations of 21 U.S.C. §§ 331(a), 331(e), 3Sl(k), and 333(a)(2) and Baldev Raj Bhutani and ALRA with violations of 18 U.S.C. §§ 2, 371, 1001, and 1341. The indictment charged all three defendants with (1) adulterating and mislabeling their Lactulose products; (2) keeping improper records with respect to the Lactulose; (3) failing to meet the good manufacturing practice standards; and (4) introducing the adulterated Lactulose into interstate commerce— all with the intent to defraud and mislead. Additionally, all three defendants were charged with adulterating and faffing to keep proper records with respect to their K+10 drug. Lastly, Baldev Raj Bhutani and ALRA were charged with providing false statements to various divisions of the Food and Drug Administration (“FDA”) and obtaining money through fraudulent schemes. While the trial delved into all of the charges against the defendants, this appeal relates only to a claim of the discovery of new, and the suppression of, evidence regarding the stability of Lactulose and the effect, if any, this had on the rest of the trial.
Lactulose is a drug that comes in a syrup form, much like the consistency of honey, that is used to treat the liver disease portal systemic encephalopathy. When an individual has this condition, the liver is unable to remove excess ammonia before it enters the bloodstream. Lactu-lose’s function is to remove the ammonia from the colon and large intestine before it can enter the bloodstream. Lactulose is a disaccharide, meaning that it is a combination of two sugars. All drugs, including Lactulose, begin a degradation process af*575ter they are manufactured, which is why they are given expiration dates. As Lac-tulose ages, it breaks down into its component parts. Should it eventually decompose fully, a process that may take several years, then it would no longer be Lactu-lose, as the two sugars will have separated, rendering the drug ineffective. Only when the two sugars are combined may they act to remove ammonia from the colon and large intestine; neither sugar on its own can remove the ammonia. Without Lactu-lose, a person with portal systemic encephalopathy could slip into a coma caused by excess ammonia in the bloodstream.
One way of testing the stability of Lac-tulose is to measure its pH level. The pH scale measures the acidity or baseness of a chemical on a scale of 0 to 14. A chemical with a pH below 7 is an acid, while one with a pH above 7 is a base. At the time of the trial, the accepted pH range in which Lactulose was considered most effective, as set forth by the U.S. Pharmacopeia (“U.S.P.”), also known as the “bible” of the pharmaceutical industry, was 3.0 to 7.0. The U.S.P. based this determination on stability data provided by various drug manufacturers to the FDA. As Lactulose ages and begins to degrade, it becomes more acidic, i.e. its pH drops. Thus, the older Lactulose gets, the lower its pH reading will become. If the pH level becomes too low, the drug will no longer be effective to fight the liver disease.
The main focus of that portion of the trial concerning ALRA’s production of Lactulose centered around two separate lots of the drug: lots 52-230 and 52-231. At trial, the government claimed, and the jury found, that employees at ALRA, at the direction of Baldev Raj Bhutani, adulterated these lots of Lactulose by “spiking” them with the foreign substance sodium hydroxide in order to conceal their age. Sodium hydroxide is a base, and, when combined with a more acidic substance, will raise its pH level. The government introduced evidence that ALRA employees opened individual bottles of Lactulose from lots 52-230 and 52-231, spiked them with sodium hydroxide, re-sealed the bottles, and repackaged them for distribution with an erroneous expiration date.
Lots 52-230 and 52-231 were manufactured in June of 1986. However, ALRA did not obtain approval from the FDA until July of 1988 to distribute the drug. When ALRA finally received approval to market its Lactulose, it aimed to distribute these two lots of Lactulose by September of 1988. The government presented evidence at trial that lot 52-231 had a pH level of 4.6 in February of 1988 and a pH level of 6.6 in August of 1988. This change represents a 100-fold increase in pH (a chemical with a pH of 4.6 is 100 times more acidic than a chemical with a pH of 6.6). In addition, the government showed that lot 52-230’s pH level increased from 4.7 to 5.1 over the same period of time. It was agreed at trial by both government and defense expert witnesses that there is no naturally occurring process which will increase the pH level of Lactulose over time.
Based on the evidence presented at trial, which consisted of expert testimony, testimony from employees of ALRA, and various data findings, on February 12, 1996, the jury found ALRA and Baldev Raj Bhu-tani guilty with respect to counts 3, 4, and 6 of the superseding indictment — charges relating to adulterating Lactulose, keeping inaccurate records, and sending the adulterated drug into interstate commerce. In total, the jury found ALRA guilty on 8 of 13 counts, Baldev Raj Bhutani on 7 of 13 counts, and Neelam Bhutani on 4 of 10 counts.
On July 5, 1996, after filing various motions for post-trial relief, the defendants filed a motion for a new trial based on newly discovered evidence and/or a Brady violation. They claimed that the government, at the time of the trial, had in its possession stability data from numerous drug manufacturing companies that showed that the effective range for Lactu-*576lose was not in fact 3.0 to 7.0, and that Lactulose was still perfectly effective with a pH level as low as 2.5. Furthermore, they asserted that the U.S.P. released a proposal in May of 1996 to change the effective pH range for Lactulose from 3.0 to 7.0 to 2.5 to 6.5. The defendants argued that, under the standards set forth in Brady, the government was obligated to turn this information over to them. They further asserted that this new data completely undermines the government’s theory of the case that the defendants were willing to put their own personal profits ahead of the safety of the public by putting adulterated drugs on the market. Additionally, they claimed that the government’s theory had a prejudicial “spillover” effect in the jury’s findings of guilty on all of the counts, even those counts unrelated to the production of Lactulose. Thus, they contend that all of the defendants are deserving of a new trial on all of their convictions.
The district court agreed with the defendants. The court stated that the government’s prevailing theme throughout the case was the defendants’ blatant disregard for public safety by shipping adulterated drugs into the marketplace just so they would not lose any profits by discarding the old, decomposed Lactulose lots. The court went on to say that since this new stability data showed that Lactulose was still effective at a lower pH level than the government represented at trial, the data tends to show that the defendants would have less of a reason to be nervous about a pH level of 4.6, and, therefore, had no need to spike the product to raise its pH level. Thus, the court determined that the government’s suppression of this new stability data “seriously undermines the court’s confidence in the verdict of the jury on Counts 3, 4 and 6.” United States v. Bhutani 1997 WL 811689 at *14 (N.D.Ill.1997). In addition, the court found that all of the counts of conviction were so entwined in the government’s profit motive theory that it ordered a new trial for all defendants on all of their convictions, even those unrelated to the Lactulose charges. The government now appeals that ruling.
II. DISCUSSION
Appellate review of a district court’s granting of a new trial in a criminal case is deferential. United States v. Boyd, 55 F.3d 239, 242 (7th Cir.1995). However, when the review of a granting of a new trial based on a Brady violation revolves around a pure issue of law, our review is de novo. United States v. Maloney, 71 F.3d 645, 652-53 (7th Cir.1995), cert. denied, 519 U.S. 927, 117 S.Ct. 295, 136 L.Ed.2d 214 (1996).
A. Brady Violation
Federal Rule 33 of Criminal Procedure allows a court to grant a defendant’s motion for a new trial if it is required in the interest of justice. A motion for a new trial based on newly discovered evidence must be made within two years after final judgment. The district court granted the defendants’ motion for a new trial because it found that the government did not fulfill its duties under Brady by not disclosing the stability data that was under the control of the FDA. We do not agree that the government committed a Brady violation.
Under Brady and its progeny, the government has an affirmative duty to disclose any evidence in its possession that is favorable to the defendant and is material to either the issue of guilt or punishment. United States v. Gonzalez, 93 F.3d 311, 315 (7th Cir.1996). Favorable evidence will be considered material if “its suppression undermines confidence in the outcome of the trial.” Id. at 316 (quoting United States v. Bagley, 473 U.S. 667, 678, 105 S.Ct. 3375, 87 L.Ed.2d 481 (1985)). The defendant does not have to show that it is more likely than not that the verdict would have been different if the evidence in question had been disclosed, but only that there is a “reasonable probability” that the outcome would have been different. Gonzalez, 93 F.3d at 316 (quoting *577Kyles v. Whitley, 514 U.S. 419, 434, 115 S.Ct. 1555, 131 L.Ed.2d 490 (1995)). Any nondisclosure of material exculpatory evidence by the government results in a violation of the defendant’s due process rights.
Even though the government has an affirmative duty to disclose exculpatory evidence in its possession, it is not obligated to disclose “every possible shred of evidence that could conceivably benefit the defendant.” United States v. Hamilton, 107 F.3d 499, 509 (7th Cir.1997), cert. denied, 521 U.S. 1127, 117 S.Ct. 2528, 138 L.Ed.2d 1028 (1997). Nor does it have an obligation to turn over evidence of which it has no knowledge. Id. Moreover, the government cannot be found to have suppressed evidence if the same information was available to the defendant through the use of reasonable diligence. United States v. Morris, 80 F.3d 1151, 1170 (7th Cir.1996), cert. denied, 519 U.S. 868, 117 S.Ct. 181, 136 L.Ed.2d 120 (1996).
While the government does not have the duty to disclose information of which it is unaware, if a government agency is charged with the administration of a statute and has consulted with the prosecution in the case, the agency will be considered part of the prosecution, and its knowledge of Brady material will be imputed to the prosecution. See United States v. Wood, 57 F.3d 733, 737 (9th Cir.1995); United States v. Anderson, 36 F.Supp.2d 1264, 1267 (D.Kan.1998). “The government cannot with its right hand say it has nothing while its left hand holds what is of value.” Wood, 57 F.3d at 737.
The material that the defendants claim was suppressed by the government is data that was in the possession of the FDA at the time of trial which said that Lactulose was perfectly stable and effective with a pH level as low as 2.5. Based on this evidence, the defendants claim that as early as late 1995, the U.S.P. planned to recommend a proposed change in the effective range of Lactulose syrup from 3.0 to 7.0 to 2.5 to 6.5. Therefore, the defendants contend that this information which was in the possession of the FDA was helpful to their case because, had they known of the proposed change, they would have informed the jury of the same and argued that the Lactulose was still effective when the alleged spiking incident occurred. The defendants claim that this evidence undermines the government’s theory of the case since the defendants would have had no reason to spike the lots in question. As it turns out, the U.S.P. eventually published its proposal to change the effective pH range of Lactulose in May of 1996, three months after the jury’s verdict.
While we agree with the defendants that information held by the FDA can, in this case, be imputed to the prosecution, we do not agree that the evidence in question is Brady material. Brady material must be (1) in the possession of the prosecution; (2) material; and (3) exculpatory. United States v. Hartbarger, 148 F.3d 777, 786 (7th Cir.1998), cert. denied, — U.S. —, 119 S.Ct. 1117, 143 L.Ed.2d 112 (1999). We disagree that the evidence in question was in the possession of the prosecution. The real evidence that the defendants rely on in their motion is the eventual publication of the U.S.P.’s proposal to lower the effective range of Lactulose. This was not published until well after the trial had ended. Simply because the FDA had stability data from other pharmaceutical companies does not mean that they had any knowledge that the U.S.P. was going to recommend the proposed change in Lactulose’s effective pH range. The government cannot be held responsible for failing to disclose merely speculative evidence. United States v. Agurs, 427 U.S. 97, 109, 96 S.Ct. 2392, 49 L.Ed.2d 342 (1976), n. 16. Since we find that the evidence in question was not within the possession of the government, we need not decide whether it was exculpatory or material. Therefore, we disagree with the district court that the defendants should have been granted a *578new trial based on the government’s alleged Brady violation.
B. Newly Discovered Evidence
Even though the evidence in question was not within the possession of the prosecution, the defendants may, nevertheless, be entitled to a new trial based on newly discovered evidence, an alternative theory raised in their motion for a new trial. The standard for determining whether a defendant is entitled to a hew trial based on newly discovered evidence is different from that of determining whether one is entitled to a new trial because of a Brady violation. We have evolved a four part test to determine whether a defendant’s motion for a new trial based on newly discovered evidence should be granted. The defendant must show that the evidence at issue (1) came to his knowledge only .after the trial; (2) could not have been discovered sooner through the use of due diligence; (3) is material and not merely cumulative or impeaching; and (4) would probably lead to an acquittal in the event of a new trial. Gonzalez, 93 F.3d at 315.
The defendants argue that the U.S.P.’s recommendation to lower the effective range of Lactulose would have been helpful to their case because it would have destroyed the prosecution’s profit-motive theory. They argue that since Lactulose does not lose any of its laxative effect and is still effective with a pH as low as 2.5, then there was no reason to spike the lots in question. The district court agreed with them. We find that both the defendants and the district court mischaracter-ized the prosecution’s argument.
The district court’s decision relies heavily on the idea that the government portrayed the defendants as profit-hungry monsters who were willing to put the public’s safety at risk in order to make a buck by putting ineffective drugs on the market. The court said that the government argued that the defendants spiked their Lactulose lots because they wanted to mask the fact that the product would soon be unfit for use. Bhutani, 1997 WL 811689 at *12 (N.D.Ill.1997). Thus, the district court concluded that the new evidence undermines the government’s profit-motive theory and the jury could very well have reached a different conclusion had it known of the new effective range.
We agree that the government’s theory was that the defendants were driven by the desire to make a profit, but, after reviewing the trial transcript, we do not agree that the government’s argument was premised on the fact that ALRA spiked the Lactulose lots because they would soon be unfit for use. In fact, the government did not argue that the Lactulose was unfit, or close to unfit. What the government did argue was that the defendants spiked the Lactulose lots because the lots were old and they wanted to conceal their age to make them look like newer lots of Lactu-lose produced by ALRA. The reason for this is that the FDA has determined that Lactulose should be marketed with an expiration date that is 18 months after its manufacture. Here, the lots in question were all manufactured in 1986 and not distributed until 1988. In order for the defendants to sell their product, it had to conform to FDA standards. Thus the defendants had to conceal the age of the product. They did this by raising the pH levels and marking the lots with expiration dates of September 1989. The government did not argue that the pH level in the lots in question ever dropped below 3.0. In fact, when the government’s expert witness, Dr. Hicks, was questioned by defense counsel, he stated that the pH levels were never outside of the effective range. Thus, it was the defense, and not the government, that put the effective range at issue in the first place. Neither Dr. Hicks, nor any other government witness, ever testified that the Lactulose in question was ineffective. The adulterated lots had the potential of becoming unsafe while in the marketplace with an inaccurate expiration date, but were not necessarily unsafe at *579the time of distribution. The government argued that the defendants put the public’s safety at risk by supplying adulterated products, not necessarily ineffective ones.
Throughout the trial, the defendants contended that the spiking incident never occurred. However, based on all of the evidence, the jury simply believed the prosecution’s case over the defendants’. The prosecution presented expert witnesses who testified that there is no naturally occurring process which can reverse the trend of Lactulose’s degradation. In addition, the government called former employees who participated in the spiking incident to describe the process that was used in adding the sodium hydroxide to the Lactulose syrup. Those employees described, in detail, the “assembly line” type procedure that was used in opening the Lactulose bottles, injecting them with sodium hydroxide -with a turkey baster type instrument and later a syringe, re-sealing the bottles, and re-packaging them with falsified expiration dates. The government also presented employee time cards with extensive overtime recorded on the days the spiking incident occurred. Lastly, these employees testified that Baldev Raj Bhutani oversaw and authorized the entire operation. The defense had ample opportunity to cross examine these witnesses, and indeed they did. The fact that the U.S.P. subsequently proposed a lower effective pH range for Lactulose is of no consequence to the case since the government never argued that the Lactulose’s pH was outside of the effective range, or even close to being outside the range.
The trial judge properly instructed the jury on what it must unanimously find in order to convict the defendants on counts 3, 4, and 6. Count 4 charged that the defendants, with the intent to defraud and mislead, failed to establish and maintain accurate drug manufacturing batch production records for the Lactulose lots in question. In order to find the defendants guilty of this count, the jury had to unanimously find that the defendants injected the lots with sodium hydroxide. The jury unanimously found this with respect to Baldev Raj Bhutani and ALRA, but not Neelam Bhutani. For count 3, the jury had to find that the lots of Lactulose were adulterated. The judge instructed the jury that the drug could have been adulterated in one of three ways. First, it could have been produced in a way that did not conform with the good manufacturing practices. Second, it could have decomposed to such a point that it no longer met applicable testing specifications. Third, the defendants could have added sodium hydroxide to it. The jury only had to find that any one of those three events occurred for the drug to be adulterated. The same is true for count 6.
Arguably, the new proposal by the U.S.P. could have shown that the Lactu-lose still met specifications even though it had degraded some. However, since the jury found in count 4 that the defendants • added sodium hydroxide to the Lactulose, then it implicitly found that the drug had been adulterated with respect to counts 3 and 6 as well. Its finding in count 4 acts as an independent basis of a finding of adulteration for counts 3 and 6. Therefore, while the new data could show that the Lactulose had not degraded to the point of making it unsafe, the jury nonetheless found that the drug was adulterated by the addition of sodium hydroxide. Thus, the defendants are not entitled to a new trial for counts 3, 4, and 6 based on newly discovered evidence.
C. The Spillover Effect
Lastly, the district court ruled that all of the defendants were entitled to a new trial on all of their convictions, even those unrelated to Lactulose, because all of the charges were so entwined with the government’s profit-motive theory that it influenced the jury’s decision on the other counts. Since we have determined that the defendants are not entitled to a new trial for counts 3, 4, and 6 because of the new evidence, it necessarily follows that *580there could be no spillover effect on the other counts. Even 'if we were to find that the defendants were entitled to a new trial for the Lactulose charges, the other convictions were based on an entirely different set of facts and conduct for a different drug altogether. These convictions were based on the manufacturing of K+10 and how a foreign substance was accidentally introduced into the mixing process, yet the drug entered the marketplace regardless. Even if the new evidence would have led to a different outcome in counts 3, 4, and 6, the new evidence has absolutely no relation to the conduct for which the defendants were found guilty in the manufacturing and record keeping of K+10. These violations contain completely different elements and are not related to the U.S.P.’s new proposal whatsoever. The jury found beyond a reasonable doubt that the defendants were guilty of these counts regardless of the Lactulose charges as evidenced by the fact that it did not find that Neelam Bhutani was guilty of counts 3, 4, and 6, but that she was guilty for her role with respect to the K+10 offenses. Therefore, the defendants are not entitled to a new trial on these additional counts.
III. CONCLUSION
For the reasons set forth above, we RevbRSe the district court’s ruling that the defendants are entitled to a new trial on all of their convictions based on a Brady violation or newly discovered evidence and Remand to the district court with orders to reinstate the verdicts of conviction and to sentence the defendants to appropriate terms as prescribed by the Sentencing Guidelines.