In these consolidated appeals, we are called upon to answer a vexing question of first impression at the appellate level: May a federal district court, consistent with the teachings of United States v. Booker, 543 U.S. 220, 125 S. Ct. 738, 160 L. Ed. 2d 621 (2005), impose a sentence outside the advisory guideline sentencing range based solely on its categorical rejection of the guidelines' disparate treatment of offenses involving crack cocaine, on the one hand, and powdered cocaine, on the other hand? The court below believed that it could and sentenced the defendants in accordance with that belief. After careful consideration, we conclude that the district court's approach was incorrect as a matter of law. Consequently, we vacate the defendants' sentences and remand for resentencing.

We start by limning the history of the disparate treatment of crack and powdered cocaine embedded in the federal sentencing guidelines (commonly referred to as the 100:1 ratio). We move from there to a brief glimpse of how that differential fared in our pre-Booker jurisprudence.

The sentencing differential for crack and powdered cocaine offenses had its genesis in the Anti-Drug Abuse Act of 1986, Pub.L. No. 99-570, 100 Stat. 3207 (1986) (codified in pertinent part at 21 U.S.C. § 841) (the Act).1 That legislation created two mandatory sentencing ranges for drug offenses. See id. § 1002 (codified at 21 U.S.C. § 841(b)(1)). The lower bracket spanned periods of imprisonment ranging from a mandatory minimum of five years to a maximum of forty years; the higher bracket spanned periods of imprisonment ranging from a mandatory minimum of ten years to a maximum of life. See id. Congress prescribed the threshold quantities of both crack and powdered cocaine needed to bring a particular offense within either bracket. See id. Despite the chemical identity of crack and powdered cocaine, Congress set widely disparate threshold quantities for the two drugs, requiring one hundred times more powdered cocaine than crack cocaine to trigger inclusion in a particular range. See id. (setting the threshold quantities for the lower range at five hundred grams of powdered cocaine and five grams of cocaine base and the threshold quantities for the higher range at five kilograms and fifty grams, respectively). Thus, for sentencing purposes, Congress treated one unit of crack on par with one hundred units of powder.2

Congress grounded this differential on its determination that crack cocaine and powdered cocaine are not fair congeners and that, all other things being equal, offenses involving the former pose a more serious societal danger than offenses involving the latter. See U.S. Sentencing Comm'n, Special Report to Congress: Cocaine and Federal Sentencing Policy 117-18 (1995) (1995 Report). In particular, Congress found that crack cocaine was more likely to (i) induce addiction; (ii) correlate with the incidence of other serious crimes; (iii) implicate especially vulnerable members of society; (iv) cause deleterious physiological effects; and (v) attract youthful users. Id. at 118.

Shortly after passage of the Act, the Sentencing Commission issued the initial compendium of federal sentencing guidelines. The Commission built the base offense levels for crimes involving crack and powdered cocaine around the threshold quantities set by Congress. This architectural decision comported with Congress's discernible intent. See 28 U.S.C. § 994(i)(5) (requiring the Commission to "specify a sentence to a substantial term of imprisonment" for offenders convicted of "trafficking in a substantial quantity of a controlled substance"). Consistent with its congressionally imposed obligation to "reduc[e] unwarranted sentence disparities," id. § 994(f), the Commission also fixed the guideline sentences for offenses involving non-threshold quantities of crack and powdered cocaine in accordance with the 100:1 ratio. See generally USSG § 2D1.1, cmt. (backg'd.) (concluding that "a logical sentencing structure for drug offenses" requires coordination with mandatory minimum sentences). Thus, while Congress designed the 100:1 ratio to operate at the minimum and maximum poles of the mandatory statutory sentencing ranges, it was the Commission that incorporated the ratio root and branch into its calculation of every cocaine offender's guideline sentencing range (GSR).

Over time, Congress began to have second thoughts about the wisdom of this dichotomy. As part of the Mandatory Minimum Reform Act of 1994, Congress enacted a safety valve provision, which provided a vehicle for lowering mandatory minimum sentences in a narrow subset of drug cases. See 18 U.S.C. § 3553(f); see also United States v. Matos, 328 F.3d 34, 38-42 (1st Cir.2003) (describing the operation of the safety valve). That same year, Congress directed the Sentencing Commission to study the crack-to-powder ratio and to submit recommendations anent its retention or modification. See Violent Crime Control and Law Enforcement Act of 1994, Pub.L. No. 103-322, § 280006, 108 Stat. 1796, 2097 (1994).

The 1995 Report embodied the Commission's response to this directive. Based on its review of the available data, it concluded that "it [could] not recommend a ratio differential as great as the current 100-to-1 quantity ratio," 1995 Report at 196, and suggested that the ratio "be re-examined and revised," id. at 197. At the same time, however, the Commission determined that empirical data supported Congress's core finding that "crack cocaine poses greater harms to society than does powder cocaine," id. at 195, and that, therefore, "important distinctions between the two may warrant higher penalties for crack," id. at xii. The Commission advised Congress that it would present more comprehensive recommendations at a future date. Id. at 198-200.

Notwithstanding its acknowledgment that higher penalties for crack offenses were justified, the Commission subsequently proposed guideline amendments designed to eliminate entirely the sentencing differential between crack and powdered cocaine. See Notice of Submission to Congress of Amendments to the Sentencing Guidelines, 60 Fed.Reg. 25,074, 25,075-25,076 (May 10, 1995). To accomplish this objective, the Commission called for reducing the base offense levels for crack offenses to the base offense levels for offenses involving equivalent quantities of powdered cocaine.

Congress held hearings on the Commission's recommendation. See 28 U.S.C. § 994(p) (providing that proposed guideline amendments must be submitted to Congress, which may "modif[y] or disapprove[ ]" them). It found that "the evidence overwhelmingly demonstrates significant distinctions between crack and powder cocaine." H.R.Rep. No. 104-272, at 3 (1995), reprinted in 1995 U.S.C.C.A.N. 335, 337. Congress also determined that a change in the guideline sentencing structure unaccompanied by a corresponding change in statutory mandatory minimums would result in "gross sentencing disparities" between offenses involving drug quantities around the threshold levels. Id. at 4. Based on these considerations, Congress rejected the Commission's proposed guideline amendments. See Pub.L. No. 104-38, § 1, 109 Stat. 334, 334 (1995).

Two years later, the Commission issued a follow-up report. See U.S. Sentencing Comm'n, Special Report to Congress: Cocaine and Federal Sentencing Policy (1997) (1997 Report). In that document, the Commission reiterated both its position that the 100:1 ratio was excessive, id. at 2, and its conclusion that "federal sentencing policy must reflect the greater dangers associated with crack," id. at 4. The Commission recommended that the 100:1 ratio be reduced to 5:1 by increasing the threshold quantities for offenses involving crack cocaine and decreasing the threshold quantities for offenses involving powdered cocaine. Id. at 2, 5, 9. Even though this report prompted the introduction of several bills aimed at reducing or eliminating the crack-to-powder disparity, Congress took no action.

In late 2001, members of the Senate Judiciary Committee invited the Sentencing Commission to update the Commission's views vis-à-vis the 100:1 ratio. The following year, the Commission issued a third report. See U.S. Sentencing Comm'n, Report to the Congress: Cocaine and Federal Sentencing Policy (2002) (2002) Report). In it, the Commission again advocated narrowing the gap that separated crack cocaine offenses from powdered cocaine offenses because (i) more recent data suggested that the penalties were disproportionate to the harms associated with the two drugs; (ii) the dangers posed by crack could be satisfactorily addressed through sentencing enhancements that would apply neutrally to all drug offenses; and (iii) the severe penalties for crack offenses seemed to fall mainly on low-level criminals and African Americans. Id. at v-viii. The Commission hastened to add, however, that crucial differences existed "in the intrinsic . . . and other harms" attributable to the two forms of cocaine and acknowledged that these differences justified stiffer penalties for crack offenses. Id. at 92. Taking into account these competing centrifugal and centripetal forces, the Commission endorsed a reduction of the 100:1 ratio to 20:1. Id. at viii. It did not, however, propose any amendments to the sentencing guidelines. Congress subsequently deliberated on the substance of the 2002 Report but did not act.

Our pre-Booker case law recognized the relative severity of the penalty paradigm for offenses involving crack cocaine but consistently deferred to Congress's policy judgments in this regard. See, e.g., United States v. Eirby, 262 F.3d 31, 41 (1st Cir.2001); United States v. Singleterry, 29 F.3d 733, 741 (1st Cir.1994). This approach produced a series of decisions that upheld the 100:1 ratio against an array of challenges. These included forays based upon the Equal Protection Clause, see, e.g., United States v. Graciani, 61 F.3d 70, 74-75 (1st Cir.1995), and the rule of lenity, see, e.g., United States v. Manzueta, 167 F.3d 92, 94 (1st Cir.1999). Pertinently, we held that, under the mandatory guidelines system that was in vogue before Booker, neither the Sentencing Commission's criticism of the 100:1 ratio nor its unrequited 1995 proposal to eliminate the differential provided a valid basis for leniency in the sentencing of crack offenders. See United States v. Andrade, 94 F.3d 9, 14-15 (1st Cir.1996) (discussing the possibility of a downward departure under USSG § 5K2.0); United States v. Sanchez, 81 F.3d 9, 11 (1st Cir.1996) (same).

It is against this backdrop that we rehearse the travel of these two cases. Because both appeals follow guilty pleas, we draw the relevant facts from each defendant's plea colloquy, the unchallenged portions of his presentence investigation report (PSI Report), and sentencing transcript. United States v. Mercedes Mercedes, 428 F.3d 355, 357 (1st Cir.2005); United States v. Dietz, 950 F.2d 50, 51 (1st Cir.1991).

On January 11, 2005, law enforcement officers searched a house occupied by defendant Sambath Pho pursuant to a warrant. They found 16.73 grams of crack cocaine together with drug-processing supplies, cash, and firearms. On June 10, 2004, Pho entered a guilty plea to a one-count information charging possession with intent to distribute five grams or more of cocaine base in violation of 21 U.S.C. § 841(a). That offense carried a mandatory minimum sentence of five years. See id. § 841(b)(1)(B).

A probation officer subsequently compiled the PSI Report. After converting the confiscated cash into its drug equivalent, see United States v. Gerante, 891 F.2d 364, 368-70 (1st Cir.1989) (explicating the provision now found at USSG § 2D1.1, cmt. n. 12), the PSI Report concluded that Pho should be held responsible for 40.43 grams of cocaine base. That yielded a base offense level of 30. A two-level gun enhancement, USSG § 2D1.1(b)(1), and a three-level discount for acceptance of responsibility, id. § 3E1.1, brought the adjusted offense level to 29. Given the absence of any meaningful prior criminal record, Pho's GSR was 87-108 months.

At the disposition hearing, the district court confirmed the adjusted offense level proposed in the PSI Report over Pho's contention that the conversion of cash was inappropriate in this instance. The court then discussed its understanding of the controlling law in a post-Booker world:

The question now is whether [the GSR] is a reasonable sentence. . . . The guidelines are no longer mandatory since Booker. The law now is that the Court impose a sentence that is reasonable whether it's in the guideline range or not, and what we're talking about is whether the guidelines produce a reasonable sentence. It is the law that the Court has to impose whatever mandatory minimum may have been prescribed by Congress, and [t]he Court cannot exceed any statutory maximum that's been prescribed by Congress. But beyond that, [the guidelines are] not the law.

After rejecting Pho's plea that his prior military service justified a lower sentence, the court described "the only real remaining question" as "this recurring question of . . . [w]hether it is fair and reasonable in calculating a sentence to consider crack cocaine as 100 times more serious than cocaine powder." (Emphasis supplied).

In the court's estimation, the 100:1 ratio was "excessive" and "not reasonable." The court explained that it had "consistently taken the position that the Commission's recommendation [of a 20:1 ratio] makes sense" and declared that, except for the 100:1 ratio, it had "no quarrel with the guideline range . . . because the guidelines take into account all the relevant factors and otherwise take an approach that is reasonable."

Based on its conclusion about the inherent unfairness of the 100:1 ratio, the court recalculated Pho's sentencing range in accordance with the 20:1 ratio. This approach produced a jerry-built sentencing range of 57-71 months. The court then reiterated:

I'm not saying that's the guideline range. The guideline range is 87 to 108 months. What I'm saying is that in order to arrive at what I would consider to be a fair and reasonable sentence that complies with the statutory criteria . . . it seems to me that it's more reasonable to use a 20 to 1 ratio, and that's how I come up with 57 to 71 months.

The court proceeded to impose an incarcerative sentence in the middle of the reconstituted range (64 months). The government filed a timely appeal. See 18 U.S.C. § 3742(b).

On September 30, 2004, police officers searched defendant Shawn Lewis's home pursuant to a warrant and discovered a virtual cornucopia of drugs: 153.75 grams of crack cocaine, 174 grams of powdered cocaine, 147 grams of marijuana, and nine grams of heroin. The search also revealed various drug-related artifacts, a large sum of cash, and two loaded .9 mm handguns. On June 3, 2005, Lewis entered a guilty plea to a two-count information charging possession with intent to distribute fifty grams or more of cocaine base and possession by a convicted felon of two handguns. See 21 U.S.C. § 841(a); 18 U.S.C. § 922(g)(1). The mandatory minimum sentence for Lewis's drug-trafficking offense was ten years. See 21 U.S.C. § 841(b)(1)(A).

A probation officer subsequently prepared the PSI Report. Based in large part on Lewis's admission that he had possessed more than one hundred fifty grams of crack cocaine, the PSI Report concluded that his base offense level for the drug-trafficking count should be 34. A two-level gun enhancement, USSG § 2D1.1(b)(1), and a three-level downward adjustment for acceptance of responsibility, id. § 3E1.1, brought the adjusted offense level to 33. Lewis's prior criminal record was extensive and placed him in the highest possible criminal history category (VI). Those calculations yielded a GSR of 235-293 months.

At the disposition hearing, the district court adopted, without objection, the findings in the PSI Report. The court then reviewed the particulars of Lewis's case. It found nothing at all in Lewis's background or upbringing that militated in favor of a shorter sentence. Moreover, the evidence indicated that he was a "fairly big time drug dealer" who had been "dealing in a significant quantity of a variety of different drugs"; that the loaded firearms posed a high degree of danger; and that he had been undeterred by previous periods of incarceration over a "long history of some serious criminal offenses." Based on this bleak picture, the court concluded that "a severe sentence is called for here."

The court then turned to Lewis's principal argument in favor of a lower sentence, namely, that the standard GSR was too punitive because it reflected the 100:1 ratio. The court described its approach to fashioning sentences in the post-Booker world:

[T]he starting point the Court always looks to is the guidelines. The guidelines generally produce a sentence that is fair under the circumstances. . . . And in this case, my starting point is to look at the guideline range and try to determine whether I think that it produces an unfair sentence and, if so, to what degree the sentence called for under the guidelines would be unfair.

Conceding that the factors delineated in 18 U.S.C. § 3553(a) supported a higher sentence than the government was recommending (the low end of the standard GSR), the court candidly acknowledged that "the only thing I see in your favor here, the only reason I would conclude that the guidelines may call for an excessive sentence, is this question . . . about the disparity between the crack cocaine and the powder cocaine." (Emphasis supplied). The court proceeded to voice its agreement with the Sentencing Commission's position that a 20:1 ratio was "more appropriate" than the 100:1 ratio because "the guidelines overstate what the penalty ought to be for crack cocaine as opposed to powder cocaine."

Starting from this premise, the court recalculated Lewis's exposure in accordance with the 20:1 ratio. This recalculation yielded a jerry-built sentencing range of 188-235 months for the drug-trafficking count. The court imposed a sentence of 188 months on that count. It also imposed a concurrent sentence of 120 months on the firearms count (which is not directly at issue in this proceeding). The government filed a timeous appeal. See 18 U.S.C. § 3742(b).

On appeal, the government's position is not that the sentences imposed were unreasonable but, rather, that the district court committed legal error. While Booker recognizes the authority of a sentencing court to tailor a sentence based on individual, case-specific considerations, the government's thesis runs, it does not give the court free rein to reject, on a categorical basis, the 100:1 ratio embedded in both the statutory scheme and the sentencing guidelines. The defendants' rejoinder is twofold. They assert that the district court's deviation from the advisory guidelines was appropriate in light of Booker and that, in all events, the district court based the sentences actually imposed on the individual circumstances of each case (not on broad-gauged policy considerations). We subdivide our analysis of this important controversy into two segments. We begin by articulating the standard of review. We then address the merits of the dispute that underlies these appeals.

It long has been the law that properly preserved challenges to a trial court's conclusions of law engender de novo review. See, e.g., United States v. Colon-Solis, 354 F.3d 101, 102 (1st Cir.2004). In the past, we have applied this standard to appellate review of a district court's interpretation of the sentencing guidelines. See, e.g., United States v. Carrasco-Mateo, 389 F.3d 239, 243 (1st Cir.2004); United States v. St. Cyr, 977 F.2d 698, 701 (1st Cir.1992).

The defendants argue that this long line of cases does not survive the remedial holding of Booker. While the defendants are correct in their observation that Booker altered the landscape of appellate review of sentencing decisions, they are incorrect in asserting that Booker displaces the de novo standard of review with respect to a sentencing court's errors of law. See United States v. Robinson, ___ F.3d ___, ___, 2005 WL 3502135 (1st Cir.2005) [No. 05-1547, slip op. at 7-8] (holding that Booker did not alter the de novo standard applicable to review of a sentencing court's legal interpretation of the guidelines). We explain briefly.

As part of its remedial holding, the Booker Court excised a provision of the Sentencing Reform Act that established standards of appellate review for certain claims of sentencing error because the operation of that provision depended on the mandatory nature of the guidelines system. See Booker, 125 S.Ct. at 765 (striking down 18 U.S.C. § 3742(e)). The Court filled the resulting lacuna with a directive that the courts of appeals thereafter should review sentences for reasonableness. Id.; see United States v. Antonakopoulos, 399 F.3d 68, 76 (1st Cir.2005) (explicating Booker).

Although the Court instructed that reasonableness review of the length of a sentence would be guided by the statutory sentencing factors, see Booker, 125 S.Ct. at 765-66, it provided no similar instruction as to how the reasonableness standard should be applied to claims — such as the one that the government presses here — that a sentencing court committed an error of law rather than an error of judgment. We agree with two of our sister circuits that, regardless of length, a sentence based on an error of law is per se unreasonable. See United States v. Price, 409 F.3d 436, 442 (D.C.Cir.2005); United States v. Crosby, 397 F.3d 103, 114 (2d Cir.2005), abrogated on other grounds by United States v. Fagans, 406 F.3d 138, 142 (2d Cir.2005).

This conclusion is reinforced by an analogy. The reasonableness standard and the familiar abuse of discretion standard bear a strong family resemblance to each other. See United States v. Ramirez-Rivera, 241 F.3d 37, 40 n. 4 (1st Cir.2001) (observing that the "practical import" of the difference between the abuse of discretion and reasonableness standards is "not immediately evident"). Under the latter standard, courts consistently have regarded an error of law as a per se abuse of discretion. See, e.g., Koon v. United States, 518 U.S. 81, 100, 116 S. Ct. 2035, 135 L. Ed. 2d 392 (1996); United States v. Snyder, 136 F.3d 65, 67 (1st Cir.1998). Accordingly, while the abuse of discretion standard contemplates substantial deference to the judgment calls of a nisi prius court, it entails "in practice . . . de novo review on issues of abstract law." Roger Edwards, LLC v. Fiddes & Son, 427 F.3d 129, 132 (1st Cir.2005). We believe that the reasonableness standard functions in precisely the same way: errors of law render a sentence per se unreasonable, and appellate review of claimed errors of law is nondeferential (i.e., de novo).

In our constitutional system, the power to define penalties for federal crimes belongs to the legislative branch of government, not the judicial branch. United States v. Evans, 333 U.S. 483, 486, 68 S. Ct. 634, 92 L. Ed. 823 (1948). While federal courts possess the discretion to tailor individual sentences within the boundaries set by the statutory framework, that discretion is subject to the limitations imposed by Congress. See Mistretta v. United States, 488 U.S. 361, 364, 109 S. Ct. 647, 102 L. Ed. 2d 714 (1989). The creation of the Sentencing Commission and the inauguration of a guideline sentencing scheme were valid exercises of congressional authority to fix penalties for federal crimes and, concomitantly, to cabin judicial discretion. See id. at 412. So too was Congress's adoption of the 100:1 crack-to-powder ratio. See Singleterry, 29 F.3d at 739-41.

Of course, Congress's authority in this area is not unbounded. Earlier this year, the Supreme Court identified a constitutional infirmity in the sentencing guidelines. See Booker, 125 S.Ct. at 756. The Booker Court held that mandatory sentencing enhancements triggered by judge-found facts were in derogation of the constitutionally assured right to trial by jury. Id. (citing U.S. Const. amend. VI). To cure that infirmity, the Court excised the statutory provision that made the sentencing guidelines binding on the federal courts. Id. at 756-57. That surgical strike rendered the guidelines effectively advisory and freed sentencing courts to tailor individual sentences in light of the factors enumerated in 18 U.S.C. § 3553(a).3 Id. at 757. This means that district courts now possess greater flexibility in reaching individual sentencing decisions.

We caution, however, that this newfound discretion, though broad, is not limitless: the guidelines remain part and parcel of the sentencing algorithm. See Booker, 125 S.Ct. at 767 (directing sentencing courts to "consult [the] Guidelines and take them into account when sentencing"); see also 18 U.S.C. § 3553(a)(4) (directing sentencing courts to consider the guidelines). It is for this very reason that the Court made it pellucid that the Sentencing Commission would continue to function and to update the guidelines with a view toward "encouraging . . . better sentencing practices. . . and promot[ing] uniformity in the sentencing process." Booker, 125 S.Ct. at 766.

Over and above the guidelines themselves, Booker recognized another significant restriction on sentencing decisions. Post-Booker, those decisions must still be grounded on the factors contained in section 3553(a). Booker, 125 S.Ct. at 766. Although the statutory sweep is wide, see supra note 3, those factors also serve to guide the discretion of sentencing courts in individual cases and thereby promote greater uniformity in sentencing decisions. See id. at 766-67.

With this mise-en-scêne in place, we turn to the propriety of the defendants' sentences. Laboring in uncharted waters, the lower court jettisoned the guidelines and constructed a new sentencing range by using a 20:1 crack-to-powder ratio in lieu of the 100:1 ratio embedded in both the statutory scheme and the guidelines. This approach, which evinced a categorical, policy-based rejection of the 100:1 ratio, amounted to error as a matter of law.

Matters of policy typically are for Congress. See, e.g., Plumley v. S. Container, Inc., 303 F.3d 364, 374 (1st Cir.2002) (explaining that "it is Congress's mission to set the policy of positive law," whereas a court's role is "to interpret that law"); United States v. Robinson, 144 F.3d 104, 110 (1st Cir.1998) (stating that the 100:1 crack-to-powder ratio is "a permissible policy choice articulated by Congress" and that, therefore, the courts "are obliged to give it effect"). A corollary to this principle is that, in the absence of constitutional infirmity, federal courts are bound by Congress's policy judgments, including judgments concerning the appropriate penalties for federal crimes. See Eirby, 262 F.3d at 41.

By congressional edict, the Sentencing Commission is allied with Congress in the important endeavor of calibrating sentences for federal offenses.4 Congress has directed the Commission to devise policies that "avoid[ ] unwarranted sentence disparities," while at the same time providing "sufficient flexibility to permit individualized sentences when warranted by mitigating or aggravating factors." 28 U.S.C. § 991(b)(1)(B) (emphasis supplied). The clear import of this statutory framework is to preserve Congress's authority over sentencing policy and to guarantee that the exercise of judicial discretion over sentencing decisions be based on case-specific circumstances, not on general, across-the-board policy considerations.

Nothing in Booker altered this distribution of authority over sentencing policy. Booker established that a district court may exercise discretion in fashioning sentences-but that discretion was meant to operate only within the ambit of the individualized factors spelled out in section 3553(a).5 See Booker, 125 S.Ct. at 764-66.

The decision to employ a 100:1 crack-to-powder ratio rather than a 20:1 ratio, a 5:1 ratio, or a 1:1 ratio is a policy judgment, pure and simple. See Andrade, 94 F.3d at 14-15 (holding that the crack-to-powder sentencing differential is not an individualized circumstance that justifies disregard of the guidelines). After all, Congress incorporated the 100:1 ratio in the statutory scheme, rejected the Sentencing Commission's 1995 proposal to rid the guidelines of it, and failed to adopt any of the Commission's subsequent recommendations for easing the differential between crack and powdered cocaine. It follows inexorably that the district court's categorical rejection of the 100:1 ratio impermissibly usurps Congress's judgment about the proper sentencing policy for cocaine offenses.

The defendants take issue with the characterization of the 100:1 ratio as a matter of congressional policy. They argue that Congress prescribed this ratio at the maximum and minimum poles of the statutory sentencing ranges but did not mandate its use in other applications. This is true as far as it goes-but it does not take the defendants very far. As the Sentencing Commission recognized when it superimposed the guidelines on the statutory framework, it would be illogical to set the maximum and minimum sentences on one construct and then to use some other, essentially antithetic construct as the basis for fashioning sentences within the range. See USSG § 2D1.1, cmt. (backg'd.). Such a patchwork sentencing scheme would run counter to Congress's goal of eliminating disparities in federal sentences. See 18 U.S.C. § 3553(a)(6); 28 U.S.C. § 994(f); see also Booker, 125 S.Ct. at 767.

Even if the defendants are correct in their assertion that the 100:1 ratio does not amount to a congressional policy choice binding on the federal courts — and we doubt that they are — the district court's categorical rejection of the 100:1 ratio still runs headlong into the will of Congress as embodied in the Sentencing Reform Act. Congress's purpose in creating a guideline sentencing scheme was to promote uniformity in federal sentences based on the "real conduct that underlies the crime of conviction." Booker, 125 S.Ct. at 759 (emphasis in original).

The district court's approach threatens to undermine this desired uniformity in two ways. In the first place, if sentencing courts are free to replace the 100:1 ratio with whatever ratio they deem appropriate, the sentences of defendants for identical "real conduct" will depend largely on which judge happens to draw a particular case. This problem has already begun to surface; in the wake of Booker, some sentencing courts have continued to impose sentences for crack offenses in lockstep with the sentencing guidelines and the 100:1 ratio while others have imposed reduced sentences based on varying ratios. Compare, e.g., United States v. Gipson, 425 F.3d 335, 337 (7th Cir.2005) (affirming sentence based on 100:1 ratio), with, e.g., United States v. Smith, 359 F. Supp. 2d 771, 782 (E.D.Wis.2005) (adopting a 20:1 ratio), and United States v. Fisher, No. S3 03 CR 1501, 2005 WL 2542916, at *6 (S.D.N.Y. Oct. 11, 2005) (adopting a 10:1 ratio).

In the second place, mandatory minimum sentences in drug cases are fixed by statute. Under the Act, a first-time offender convicted of an offense involving fifty or more grams of crack cocaine is subject to a ten-year mandatory minimum sentence. See 21 U.S.C. § 841(b)(1)(A). Because the current guidelines were constructed around the mandatory minimums, a first-time offender convicted of an offense involving, say, forty-nine grams of crack would likely encounter a GSR of 97-121 months.6 See USSG § 2D1.1(c)(5); id. Ch.5, Pt.A. If, however, a sentencing court displaced the guidelines by applying a 20:1 crack-to-powder ratio, the second offender's GSR would plummet to 63-78 months. See 2002 Report, Appx. at A-4; USSG Ch.5, Pt.A. In short, a one-gram difference in drug quantity would create a huge sentencing differential (nearly fifty percent). It was exactly this sort of concern that drove Congress's decision to reject the proposed guideline amendments in 1995. See H.R.Rep. No. 104-272, supra, at 4, reprinted in 1995 U.S.C.C.A.N. at 337.

The defendants concede that the categorical adoption of a 20:1 ratio by a single district judge would create these distorting effects. They nonetheless argue that such a praxis reduces sentencing disparities. This is so, the defendants aver, because the 20:1 ratio places sentences for crack and powdered cocaine offenses on a more even keel. This argument misapprehends the nature of the disparity that ought to concern us.

Although the district court's approach does ameliorate the disparity in sentences for crack and powdered cocaine offenses, what counts is the uniformity in sentencing sought by Congress. That uniformity "does not consist simply of similar sentences for those convicted of violations of the same statute" but "consists, more importantly, of similar relationships between sentences and real conduct." Booker, 125 S.Ct. at 761. Congress plainly believed that not all cocaine offenses are equal and that trafficking in crack involves different real conduct than trafficking in powder. Otherwise, it would not have ordered dissimilar treatment of the two types of offenses in the Act. Clearly, then, Congress intended that particular disparity to exist, and federal courts are not free to second-guess that type of decision. Cf. Snyder, 136 F.3d at 68-70 (rejecting argument that disparity between federal and state sentences for similar firearms offenses justified reduced sentence because Congress deliberately created the disparity through the Armed Career Criminal Act).

In an effort to sidestep the force of this reasoning, the defendants assert that the sentences in these cases resulted from individualized analysis rather than from any categorical imperative. For example, Lewis points out that, at his sentencing, the lower court addressed a number of the section 3553(a) factors. In a similar vein, Pho alludes to the district court's discussion of several particularized considerations. We find this argument unpersuasive.

The sockdolager is that, although the district court considered individualized factors in each case, it did not base its sentencing decisions on those factors. By the district court's own description, "the only thing" that supported reduced sentences for the defendants was the "tragic" and "incongruous" treatment of crack cocaine offenses in the sentencing guidelines and the underlying statutory scheme. The court made no bones about its intention to apply its preferred 20:1 ratio categorically in future cases:

I've encountered this in other cases, and I've consistently taken the position that the [20:1 ratio] makes sense. . . . I would imagine that by now this question must be under consideration by the First Circuit . . . and I hope we get some clarification on this sometime soon before we build up too much of a backlog of these cases. But that's the position I take now.

Thus, the record, fairly read, belies the insinuation that the district court imposed the sentences appealed from on the basis of case-specific factors.

To recapitulate, we hold that the district court erred as a matter of law when it constructed a new sentencing range based on the categorical substitution of a 20:1 crack-to-powder ratio for the 100:1 ratio embedded in the sentencing guidelines. This holding recognizes that sentencing decisions must be done case by case and must be grounded in case-specific considerations, not in general disagreement with broad-based policies enunciated by Congress or the Commission, as its agent.

Let us be perfectly clear. We do not intend to disparage the district court's thoughtful attempt to deal with a problem that has tormented many enlightened observers ever since Congress promulgated the 100:1 ratio. By the same token, we do not intend to diminish the discretion that, after Booker, district courts enjoy in sentencing matters or to suggest that, in a drug-trafficking case, the nature of the contraband and/or the severity of a projected guideline sentence may not be taken into account on a case-by-case basis. Our goal is simply to channel the district courts' newfound discretion in ways that both comport with the Booker Court's remedial opinion and respect the separation of powers between the legislative and judicial branches of government. While we share the district court's concern about the fairness of maintaining the across-the-board sentencing gap associated with the 100:1 crack-to-powder ratio, the proper place to assuage that concern is in the halls of Congress, not in federal courtrooms. In the final analysis, it is Congress, not the courts, that possesses the institutional capacity to address the problem in a coherent and uniform fashion. See Singleterry, 29 F.3d at 741.

We need go no further. For the reasons elucidated above, we vacate the defendants' sentences and remand for resentencing in accordance with this opinion. We intimate no view as to the length of the sentences to be imposed on remand.

Vacated and remanded.