CALABRESI, Circuit Judge.

The government appeals from the judgment of the district court (Sweet, /.), dismissing the indictments against the defendants on the ground that the definition of “controlled substance analogue,” 21 U.S.C. § 802(32)(A), as applied to 1,4-butanediol, is unconstitutionally vague. We conclude that, because 1,4-butanediol both (a) differs from gamma hydroxybutyric acid (“GHB”), a Schedule I controlled substance, by only two atoms and (b) converts into GHB upon ingestion, it is sufficiently clear that 1,4-butanediol is “substantially similar [in] chemical structure” to GHB, as that term is used in § 802(32)(A)’s definition of “controlled substance analogue,” and, hence, that the definition as applied in this case does not violate the Constitution. The defendants do not contest that 1,4-butanediol meets the other statutory criteria for being a controlled substance analogue. We therefore vacate the dismissal of the indictments and remand for further proceedings.

BACKGROUND

On April 26, 2001, a grand jury in the Southern District of New York returned *120 indictments against Todd Kelly Roberts and Michael Toback, charging that Roberts and Toback conspired to distribute and distributed 1,4-butanediol for human consumption, in violation of 21 U.S.C. §§ 802(32)(A), 812, 813, 841(a)(1), 841(b)(1)(C), and 846. The government charges that Toback made retail sales of 1,4-butanediol from a New York City health food store. It also charges that Toback obtained the 1,4-butanediol from the Barin Corporation and that Roberts is the president of that company.

1,4-butanediol has a long history of use as an industrial solvent. In recent years it has been discovered to be a depressant, inducing a deep sleep. According to the government, this property has caused it to be used as a “date rape” drug. Though 1,4-butanediol has not been listed as a controlled substance, see 21 U.S.C. § 812; 21 C.F.R. §§ 1308.11-1308.15, the government contends that it is a “controlled substance analogue” of GHB, which Congress listed as a Schedule I controlled substance in 2000. See Hillory J. Farias and Samantha Reid Date-Rape Drug Prohibition Act of 2000, Pub.L. No. 106-172, § 3(a)(1), 114 Stat. 7 (2000) (“Pub.L. No. 106-172”). Under the Controlled Substance Analogue Enforcement Act of 1986 (the “Act”), 21 U.S.C. §§ 802(32)(A), 813 (2000), some substances that have not been listed as controlled substances, but that are sufficiently similar to a listed substance, are, under certain additional circumstances, treated as Schedule I controlled substances for the purposes of federal drug law.

After their indictment, the defendants submitted Rule 12(b) motions to dismiss. See Fed.R.Crim.P. 12(b). In these, they argued, inter alia, that the Act was so vague (as applied to 1,4-butanediol) that the defendants’ prosecution violated the Fifth Amendment’s guarantee of due process. On December 14, 2001, the district court (Sweet, J.) granted a hearing on the matter and issued a preliminary ruling on how the Act’s definition of “controlled substance analogue” should be interpreted.

The Act defines a “controlled substance analogue” as a substance:

(i) the chemical structure of which is substantially similar to the chemical structure of a controlled substance in schedule I or II;

(ii) which has a stimulant, depressant, or hallucinogenic effect on the central nervous system that is substantially similar to or greater than the stimulant, depressant, or hallucinogenic effect on the central nervous system of a controlled substance in schedule I or II; or

(iii) with respect to a particular person, which such person represents or intends to have a stimulant, depressant, or hallucinogenic effect on the central nervous system that is substantially similar to or greater than the stimulant, depressant, or hallucinogenic effect on the central nervous system of a controlled substance in schedule I or II.

21 U.S.C. § 802(32)(A). The interpretive question that the district court addressed in its December 14, 2001 order was whether the three subsections of the definition should be read in the disjunctive or in the conjunctive. Under the disjunctive reading, a substance that satisfies one or more of the subsections is a “controlled substance analogue.” According to the conjunctive reading, the definition requires two things: first, (i) that the substance be chemically similar and, second, (ii) that it have a similar or greater psychopharmaco-logieal effect or (iii) that it be intended to have or be represented as having such an effect. Before the district court, the government argued for the disjunctive reading, the defendants for the conjunctive.

*121 The district court found that the language of the statute did not unambiguously lend itself to either reading, because the “which” in (ii) and (iii) appears at the beginning of those clauses, suggesting that it might refer not to “substance” in .the preface of the definition, but instead to “chemical structure” in (i). Having found the provision ambiguous, the district court examined the purpose and legislative history of the Act and concluded that the conjunctive reading was the correct one. Other courts that have considered the question have come to the same conclusion. United States v. Hodge, 321 F.3d 429, 432-39 (3d Cir.2003) (providing an extensive analysis of the plain meaning and legislative history of 21 U.S.C. § 802(32)(A)); United States v. Washam, 312 F.3d 926, 930 n. 2 (8th Cir.2002) (adopting the conjunctive reading); United States v. McKinney, 79 F.3d 105, 107-08 (8th Cir.1996), vacated on other grounds, 520 U.S. 1226, 117 S.Ct. 1816, 137 L.Ed.2d 1025 (1997) (paraphrasing the test in the conjunctive without discussion); United States v. Forbes, 806 F.Supp. 232, 236 (D.Colo.1992) (stating that the conjunctive reading is required to avoid unintended results and is supported by the Act’s legislative history). But see United States v. Fisher, 289 F.3d 1329, 1338 (11th Cir.2002) (expressly declining to decide whether the conjunctive or disjunctive reading governs); United States v. Granberry, 916 F.2d 1008, 1010 (5th Cir.1990) (stating the test in the disjunctive without discussion). Because the government has not appealed this holding of the district court, we assume that the conjunctive reading is correct.

On June 17, 2002, the district court held a hearing on the defendants’ void-for-vagueness argument. The defendants did not contest that 1,4-butanediol satisfied clause (ii) of the definition of “controlled substance analogue” — which requires a stimulant, depressant, or hallucinogenic effect on the central nervous system that is similar to or greater than that of a controlled substance. At issue was only whether it was sufficiently clear that 1,4-butanediol satisfied clause (i) of that definition' — requiring a “chemical structure [that is], substantially similar to the chemical structure of a controlled substance.”

Two defense experts, Drs. Haley and Schuster, and one government expert, Dr. DiBerardino, testified at the June 17, 2002 hearing. Haley and Schuster are both professors of chemistry. DiBerardino has a degree in polymer chemistry and is employed by the DEA. The experts agreed on the following facts as to the chemical makeup of 1,4-butanediol and GHB. (1) The molecules of 1,4-butanediol and GHB differ by two atoms: an oxygen atom in GHB is replaced by two hydrogen atoms in 1.4-butanediol. (2) The different atoms are located in a “functional group” of the molecule, which is a section of the molecule that is especially important in determining the substance’s properties. In this case, the difference means that 1,4-butanediol is an alcohol while GHB is an acid and that the chemicals have different reactivity, acidity levels, and melting and boiling points. (3) While the difference between 1.4-butanediol and GHB appears minor on a two-dimensional chart, the molecules have different three-dimensional structures. The four-rcarbon chain.in 1,4-buta-nediol is linear, while in GHB the chain folds ,oyer itself. (4) The nuclear magnetic resonance (NMR) spectrograms of 1,4-bu-tanediol and GHB are significantly different. (5) Despite the difference in functional- groups, upon ingestion, 1,4-butanediol is metabolized, in two steps, into GHB.

The experts, however, disagreed on whether 1,4-butanediol was “substantially similar [in] chemical structure” to GHB. The defense experts argued that 1,4-buta- *122 nediol and GHB were not substantially similar in chemical structure for the following reasons: (1) chemists treat alcohols and acids as belonging to fundamentally different categories; (2) 1,4-butanediol is linear, while GHB folds over on itself; and (3) chemists commonly use NMR spectro-grams to classify chemicals and the NMR spectrograms of 1,4-butanediol and GHB are significantly different. The prosecution expert, DiBerardino, argued that functional groups concerned a molecule’s various properties and reactivity, but not its structure, and emphasized the similarity of the two-dimensional diagrams.

On September 9, 2002, the district court issued an opinion granting the defendants’ 12(b) motions to dismiss the indictments. In order to evaluate the defendants’ as-applied void-for-vagueness challenge, the district court used the two-part test described in Kolender v. Lawson, 461 U.S. 352, 103 S.Ct. 1855, 75 L.Ed.2d 903 (1983). To satisfy the Fifth Amendment’s due process requirements, a penal statute must “define the criminal offense [1] with sufficient definiteness that ordinary people can understand what conduct is prohibited and [2] in a manner that does not encourage arbitrary and discriminatory enforcement.” Id. at 357, 103 S.Ct. 1855.

With respect to the first prong, adequate notice, the court held that because the definition of “controlled substance analogue” is a scientific one, the terms in that definition should be explained by reference to the field of science to which they relate. The court therefore focused on the lack of consensus among the experts in the June 17 hearing and reasoned that if the scientific community could not agree on the proper methodology to determine structural similarity, a reasonable layperson would not be able to ascertain what the term meant as applied to 1,4-butanediol.

The district court also held that, as applied to 1,4-butanediol, the Act failed the second prong of the void-for-vagueness test, because it did not provide adequate enforcement guidelines. The court took as evidence of the danger of arbitrary enforcement the fact that there were ten naturally occurring substances that were also chemically related to GHB — some differing from it by only one functional group — none of which had been subject to prosecution by the government. It further reasoned that the Act’s only apparent limitation on prosecution was its focus on “designer drugs” — -i.e., drugs specifically developed to mimic controlled substances— and that this limitation did not apply to 1,4-butanediol, since it had existed and had been openly sold long before its problematic uses were identified.

DISCUSSION

Because a motion to dismiss on grounds that a criminal statute provides insufficient notice raises a question of law, we review the dismissal of the indictment by the district court de novo. United States v. Velastegui, 199 F.3d 590, 593 (2d Cir.1999).

The district court correctly held that defendants’ constitutional challenge is governed by Kolender, under which “a court must first determine whether the statute gives the person of ordinary intelligence a reasonable opportunity to know what is prohibited and then consider whether the law provides explicit standards for those who apply it.” Chatin v. Coombe, 186 F.3d 82, 87 (2d Cir.1999) (citation omitted). The first prong of the Kolender test is based on the principle that no one “shall be held criminally responsible for conduct which he could not reasonably understand to be proscribed.” United States v. Harriss, 347 U.S. 612, 617, 74 S.Ct. 808, 98 L.Ed. 989 (1954). The Supreme Court has *123 unpacked Kolender’s adequate-notice requirement as follows:

There are three related manifestations of the fair warning requirement. First, the vagueness doctrine bars enforcement of a statute which either forbids or requires the doing of an act in terms so vague that men of common intelligence must necessarily guess at its meaning and differ as to its application. Second, as a sort of junior version of the vagueness doctrine, the canon of strict construction of criminal statutes, or rule of lenity, ensures fair warning by so resolving ambiguity in a criminal statute as to apply it only to conduct clearly covered. Third, although clarity at the requisite level may be supplied by judicial gloss on an otherwise uncertain statute, due process bars courts from applying a novel construction of a criminal statute to conduct that neither the statute nor any prior judicial decision, has fairly disclosed to be within its scope. In each of these guises, the touchstone is whether the statute, either standing alone or as construed, made it reasonably clear at the relevant time that the defendant’s conduct was criminal.

United States v. Lanier, 520 U.S. 259, 266-67, 117 S.Ct. 1219, 137 L.Ed.2d 432 (1997) (internal quotation marks and citations omitted). Kolender’s second prong requires that legislatures “establish minimal guidelines to govern law enforcement” in order to prevent “policemen, prosecutors, and juries [from] pursuing] their personal predilections.” Smith v. Goguen, 415 U.S. 566, 574-75, 94 S.Ct. 1242, 39 L.Ed.2d 605 (1974).

Because the statute at issue here contains a scienter requirement, see 21 U.S.C. § 841(a), the defendants’ vagueness challenge must be met with some measure of skepticism, at least with regard to the “fair notice” prong of Kolender. 1 See Village of Hoffman Estates v. Flipside, Hoffman Estates, Inc., 455 U.S. 489, 499, 102 S.Ct. 1186, 71 L.Ed.2d 362 (1982) (“[A] scienter requirement [in a criminal statute] may mitigate a law’s vagueness, especially with respect to the adequacy of notice to the complainant that his conduct is proscribed.’’). That skepticism is further buttressed by the government’s proffer that the defendants actually believed what they were doing was illegal. See Washam, 312 F.3d at 930; United States v. Saffo, 227 F.3d 1260, 1270 (10th Cir.2000). But even putting aside these considerations, the defendants’ challenge fails on its own terms. See United States v. Rybicki, 354 F.3d 124 (2d Cir.2003) (recognizing that 18 U.S.C. § 1346, read in conjunction with §§ 1341 and 1343, contains a scienter requirement, but analyzing whether the term “honest services” is sufficiently, definite to survive Kolender’s first prong).

The defendants argue that, given Kolender’s requirements, the phrase “substantially similar [in] chemical structure” is unacceptably vague as applied to 1,4-buta-nediol. In response, the government re *124 lies (in part) on discussions of 1,4-butane-diol that appear in congressional findings and Drug Enforcement Agency (DEA) regulations, which it contends sufficed to put the defendants on constructive notice that 1,4-butanediol was a controlled substance analogue. 2 It might be argued, however, that the Act gives neither Congress nor the DEA authority to determine that a chemical is a controlled substance analogue. Instead, it leaves that decision to the courts and, as a result, in the absence of prior court decisions the statutory and regulatory pronouncements provide no real notice. 3 We do not, however, need to choose between these positions, because the government makes two additional arguments that, though perhaps individually unsuccessful, when combined (as they were at oral argument) yield a definition of “substantially similar [in] chemical structure” that is both supported by the language and purpose of the statute and quite sufficient to avoid vagueness as applied to the facts of the case before us.

The government asserts, first, that the visual similarity between the two-dimensional diagrams of the molecules of 1,4-butanediol and GHB — which differ from each other by only two atoms — is, standing alone, enough to establish “substantial similarity [in] chemical structure.” We doubt, however, that this is an appropriate rule to apply in every situation. In another case, it might well be that a one- or two-atom difference in a molecule made such a radical difference in the substance’s relevant characteristics that any similarity in two-dimensional charts would not be “substantial” enough to satisfy the definition of “controlled substance analogue.” As a result, we decline to adopt the government’s proposed test.

As the government also emphasizes, however, 1,4-butanediol is metabolized into GHB after ingestion. Again, the government’s brief seeks to make too much of this fact, for it argues that such post-ingestion conversion suffices, on its own, to establish that 1,4-butanediol is substantially similar in chemical structure to GHB. Cf. United States v. Fisher, 289 F.3d 1329, 1339 (11th Cir.2002) (“People of ordi *125 nary intelligence would easily be able to determine that a substance, which is converted upon ingestion into a metabolite with a substantially similar chemical structure and effect on the central nervous system as a schedule I controlled substance, would meet the definition of a controlled substance analogue.”)

We are uncertain whether in a case where, prior to ingestion, there is no readily cognizable chemical similarity between the suspect substance and a controlled substance (e.g., where there is no visual similarity between their two-dimensional diagrams), the fact that, once in the body, the suspect substance converts into a controlled substance is enough to establish substantial similarity in chemical structure under the Act. As the defendants contend, if the chemical structure of 1,4-butanediol were to be determined only after ingestion, then 1,4-butanediol would not be a controlled substance analogue, but would be a controlled substance. This is because after ingestion the substance is not similar to GHB; it is identical to it. Moreover, as the defendants further assert, such a definition of “analogue” would offend the plain language of the Act and if this unusual meaning were what Congress intended by “substantially similar [in] chemical structure,” then the Act would not put the public on sufficiently clear notice that the distribution of 1,4-butanediol was illegal. Whatever their ultimate merits, these arguments cannot be dismissed off-hand. We therefore decline, in a case where it is unnecessary to the outcome, to consider further the government’s proposed broad rule.

But, as the government pressed at oral argument, this is a case where there is both readily cognizable similarity prior to ingestion and metabolizatiqn into the controlled substance after ingestion. Where there is only a two-atom difference between the relatively complex molecules of a suspect substance and of a controlled substance and where, upon ingestion, the suspect substance is metabolized into the controlled substance, we believe that the chemical structure of the suspect substance is manifestly “substantially similar to the chemical structure of [the] controlled substance,” as that phrase is used in the definition of “controlled substance analogue.” And this is so whether one focuses on the language or on the purpose of the statute. 4 Consequently, a jury could reasonably find that 1,4-butanediol, which differs from GHB by only two atoms and is converted into GHB upon ingestion, easily meets the requirements of clause (i) of the definition of “controlled substance analogue.” 5 We conclude that the defen *126 dants, who knew or could have readily ascertained these facts, had sufficient warning that 1,4-butanediol was a controlled substance analogue and that its distribution for human consumption was illegal.

We also believe that the aforementioned meaning of “controlled substance analogue” is definite enough with respect to 1,4-butanediol to meet the second prong of the Kolender test, i.e., the requirement that a criminal offense be defined “in a manner that does not encourage arbitrary and discriminatory enforcement.” 461 U.S. at 357, 103 S.Ct. 1855.

The district court gave two reasons for its finding of vagueness in this respect. The first was that there are ten other, naturally occurring substances that, in the court’s view, were chemically similar to GHB. Several of these, the court said, were comparable to 1,4-butanediol in that they differed from GHB by one functional group. Yet none of them had been the subject of government prosecution. The court’s second reason was that unless limited to “designer drugs” — substances synthesized to mimic the properties of a controlled substance — the Act would not provide sufficient guidelines for enforcement. We find neither argument convincing.

As to the government’s failure to prosecute the sale of the ten naturally occurring substances that are allegedly chemically similar to GHB, the district court did not find, and the defendants have not argued, that those substances metabolize into GHB. Nor have they asserted that the substances meet the second or third prong of the definition of “controlled substance analogue” — that they, like 1,4-butanediol, have a stimulant, depressant, or hallueino-genic effect on the central nervous system that is similar to or greater than that of GHB or that they have been represented or intended to have such an effect. It appears quite possible, therefore, that the reason these substances have not been the subject of prosecution is that they do not satisfy the definition of “controlled substance analogue,” either to the extent that clause (i) requires metabolization into a controlled substance, or because that definition is further limited by clauses (ii) and (iii). 6

With respect to the fact that 1,4-butane-diol is not a “designer drug,” it is true that Congress seems to have been first prompted to address the issue of controlled substance analogues by the perceived problem of designer drugs. But the wording of the Act evinces a clear congressional intent not to limit its scope to such substances. Moreover, the Act specifically protects against possible arbitrary enforcement where a controlled substance analogue has a pre-existing, non-pharmacological use by stipulating that a controlled substance analogue is to be treated as a Schedule I controlled substance only “to the extent intended for human consumption.” 21 U.S.C. § 813. This provision places sufficient additional restraints on the enforcement of the Act so that, at least with respect to the facts before us, the guidelines for enforcement satisfy the Fifth Amendment’s guarantee of due process.

The district court, in its finding of vagueness, put great weight on the seeming lack of consensus among the experts it examined in the June 17 hearing. It may be that, in a case in which it was not obvious whether the chemical clearly fell within the intended meaning of “controlled *127 substance analogue,” such disagreement among experts might be relevant. But that is not the case before us. Given the combination of GHB’s cognizable similarity to 1,4-butanediol prior to ingestion and its metabolization into that controlled substance after ingestion, the classification of 1,4-butanediol as a controlled substance analogue is clearly mandated by the Act’s language, and remains so regardless of the differences of view among the experts. In short, given the circumstances of this case, an evaluation of the statute’s vagueness as-applied does not call for the fine distinctions drawn by the experts.

We conclude that the Act’s definition of “controlled substance analogue” gives the public clear enough warning that 1,4-buta-nediol qualifies as a controlled substance analogue and sufficiently limits prosecuto-rial and judicial discretion in its enforcement, and that, therefore, the definition of “controlled substance analogue” is not unconstitutionally vague as applied to 1,4-butanediol. Accordingly, we VACATE the judgment of the district court and Remand for further proceedings consistent with this opinion.

Notes

. Under the statutory scheme here, the defendants cannot be convicted unless a jury concludes beyond a reasonable doubt that (1) they possessed 1,4-butanediol, (2) which was similar in structure and effect to a listed controlled substance, (3) which was intended for human consumption (bringing it within the definition of a controlled substance), (4) with the intent to distribute it, and (5) they did so with the knowledge that they were in possession of a controlled substance. 21 U.S.C. §§ 802(32)(A), 813, 841(a)(1). The last element requires the Government to show that the defendants knew that they possessed a controlled substance, not that they "might be involved in some sort of criminal activity.” United States v. Morales, 577 F.2d 769, 773 (2d Cir.1978). The defendants, however, need not know the exact nature of the drug; it is sufficient that they be aware that they possessed "some controlled substance.” Id. at 776.

. See Pub.L. No. 106-172, § 2(4) (2000) ("If taken for human consumption, common industrial chemicals such as gamma butyrolac-tone and 1.4-butanediol are swiftly converted by the body into GHB. Illicit use of these and other GHB analogues and precursor chemicals is a significant and growing law enforcement problem.”); Placement of Gamma-Butyrolactone in List I of the Controlled Substances Act, 65 Fed.Reg. 21645, 21645 (Apr. 24, 2000) (codified at 21 C.F.R. § 1310.02) ("If taken for human consumption, GBL and other chemicals, including 1,4-butanediol, are swiftly converted into GHB by the body. Abuse of these and other GHB-like substances is a significant law enforcement and public health problem. GBL and 1,4-butanediol are structurally and phar-macologically similar to GHB and are often substituted for GHB. Under certain circumstances they may satisfy the definition of a controlled substance analogue (21 U.S.C. 802(32)).”); Addition of Gamma-Hydroxybu-tric Acid to Schedule I, 65 Fed.Reg. 13235, 13236 (Mar. 13, 2000) ("[The] FDA has also declared 1,4-butanediol, a chemical related to both GHB and GBL, a Class I Health Hazard. On May 11, 1999, the FDA issued another warning on 1,4 butanediol, GHB, and GBL stating that these substances pose a significant health hazard.”)

. It is perhaps unfortunate that Congress did not opt to list known controlled substance analogues itself, and then to delegate to an appropriate designee (like the Attorney General) the authority to expand that list as necessary, but rather left the determination of what qualifies as a controlled substance analogue to the courts and to informal legislative or administrative commentary. Legislative or administrative listing would have automatically put the public on clear notice that those chemicals — to the extent they were intended for human consumption — would be treated for the purposes of federal law as a Schedule I controlled substance. See 21 U.S.C. § 813.

. We emphasize that the fact that 1,4-butane-diol metabolizes into GHB is significant not because it means that the chemicals are identical, but because it indicates that the readily ascertainable pre-ingestion similarity between 1,4-butanediol and GHB — the mere two-atom difference between the molecules — is a similarity of a material and relevant sort, given the purpose of the Act.

. The district court suggested that to treat the fact that 1,4-butanediol converts in the body into GHB as relevant to clause (i) would be to conflate clause (i) with clause (ii). But clause (ii), which requires an effect "substantially similar to or greater than the [controlled substance’s] stimulant, depressant, or hallucinogenic effect on the central nervous system,” concerns not the chemical properties of the substance in question (e.g., what it converts into), but its psychopharmacological effect. A chemical might be metabolized into a controlled substance and yet not have a similar or greater stimulant, depressant, or hallucinogenic effect on the central nervous system. This might be so, for instance, because the metabolized controlled substance appeared in much smaller quantities or because other blocking agents were present. Considering a substance’s metabolites as relevant to its chemical structure, therefore, does not conflate the two prongs of the definition:

. We note that the notion that clauses (ii) and (iii) place additional, constitutionally relevant limits on prosecutorial discretion would not apply were the disjunctive reading of the Act’s definition of "controlled substance analogue” accepted as correct.