We must decide whether Title I of the Americans with Disabilities Act, 42 U.S.C. sec.sec. 12111-17, is an exercise of power under sec.5 of the Fourteenth Amendment, which confers authority "to enforce, by appropriate legislation, the provisions of this article." Defendant in this suit is an arm of Illinois and therefore one of the United States for purposes of the Eleventh Amendment. Congress has power under the Commerce Clause to adopt the ADA's rules, but given the Eleventh Amendment a statute that rests only on the Commerce Clause can not authorize private suits against states in federal court. Seminole Tribe v. Florida, 517 U.S. 44 (1996). But if sec.5 bestows power to adopt the ADA, then private litigation is compatible with the Eleventh Amendment. Fitzpatrick v. Bitzer, 427 U.S. 445 (1976).

Melinda Erickson worked for five years in the College of Business and Management at Northeastern Illinois University, rising from secretary to "program associate." She contends that the University failed to accommodate her efforts to have children. Medical care for her infertility was physically demanding and had side effects. Both the treatment and the circumstances that gave rise to it were emotionally draining. Erickson often did not come to work and was late on days when she did appear. She was fired after she became distraught and stayed home for six working days. Erickson does not contend that the attendance requirements were designed to discriminate against persons with disabilities. Instead she argues that the University should have tolerated absences and tardiness that it would not have condoned from a healthy employee. Invoking the Eleventh Amendment, the University filed a motion to dismiss, which the district court denied. 1998 U.S. Dist. Lexis 15779 (N.D. Ill. Oct. 1, 1998). The University's interlocutory appeal is within our jurisdiction, see Seminole Tribe, 517 U.S. at 52, even though the University does not assert sovereign immunity with respect to Erickson's claim under the Pregnancy Discrimination Act, 42 U.S.C. sec.2000e(k). Scott v. Lacy, 811 F.2d 1153 (7th Cir. 1987). Cf. Wisconsin Department of Corrections v. Schacht, 524 U.S. 381 (1998). The United States intervened as a party in this court to defend the ADA's constitutionality. See 28 U.S.C. sec.2403(a).

Three times during the last four Terms, the Supreme Court has addressed the extent of legislative power under sec.5. Kimel v. Florida Board of Regents, 120 S. Ct. 631 (2000); Florida Prepaid Postsecondary Education Expense Board v. College Savings Bank, 527 U.S. 627 (1999); Boerne v. Flores, 521 U.S. 507 (1997). Thrice it has stressed that the language of sec.5, which gives Congress the power to "enforce" the Fourteenth Amendment, must be taken seriously. Statutes that create new rights, or expand old rights beyond the Fourteenth Amendment's bounds, do not "enforce" that amendment.

Boerne dealt with the Religious Freedom Restoration Act of 1993 (RFRA), 42 U.S.C. sec.sec. 2000bb to 2000bb-4, a response to Employment Division v. Smith, 494 U.S. 872 (1990). Smith had held that the Free Exercise Clause of the First Amendment never requires accommodation of religiously inspired practices, so that laws neutral with respect to religion are valid. The RFRA, by contrast, obliged states to accommodate practices associated with religion. The Court held that an accommodation requirement could not be thought to "enforce" a constitutional norm that does not require accommodation. Florida Prepaid held that Congress may not use sec.5 to abrogate state sovereign immunity on the ground that statutory rights are "property" under the Fourteenth Amendment. Kimel held that sec.5 does not support the Age Discrimination in Employment Act, 29 U.S.C. sec.sec. 621-34, because although the ADEA forbids consideration of an employee's age unless age is a "bona fide occupational qualification reasonably necessary to the normal operation of the particular business", sec.623(f)(1), the Constitution's own requirement is considerably more lenient. The Equal Protection Clause permits a state to consider a person's age unless age lacks a rational relationship to the state's objective. Most consideration of age in employment therefore is constitutional; but under the ADEA most consideration of age is forbidden; Kimel therefore held that the ADEA sets up an independent rule and does not "enforce" the Constitution's rule.

Twenty-three days before the Supreme Court decided Boerne, we held in Crawford v. Indiana Department of Corrections, 115 F.3d 481, 487 (7th Cir. 1997), that sec.5 supports Title II of the ADA, which deals with public services. Our opinion analogized the ADA to the ADEA and observed that the latter statute had been applied to states in private litigation. Kimel shows that if our analogy to the ADEA is precise, then Crawford is no longer authoritative; Florida Prepaid and Boerne likewise call for a fresh look at the subject. Elsewhere a great deal of ink has been spilled on this question. After Boerne but before Kimel, panels of five appellate courts held that sec.5 supplies the necessary legislative power, though there was one squarely contrary holding by a court en banc. Compare Muller v. Costello, 187 F.3d 298 (2d Cir. 1999); Coolbaugh v. Louisiana, 136 F.3d 430 (5th Cir. 1998); Clark v. California, 123 F.3d 1267 (9th Cir. 1997); Martin v. Kansas, 190 F.3d 1120 (10th Cir. 1999); and Kimel v. Florida Board of Regents, 139 F.3d 1426, 1433, 1441-44 (11th Cir. 1998), with Alsbrook v. Maumelle, 184 F.3d 999 (8th Cir. 1999) (en banc). The fourth circuit is internally divided. Although Amos v. Maryland Department of Public Safety, 178 F.3d 212 (4th Cir. 1999) (rehearing en banc granted Dec. 28, 1999), holds that private ADA litigation may proceed against state prisons, Brown v. North Carolina Division of Motor Vehicles, 166 F.3d 698 (4th Cir. 1999), held that a regulation, based on the ADA, requiring the state's Division of Motor Vehicles to accommodate disabled drivers, is unconstitutional. Recently a divided panel of the ninth circuit disagreed with Brown. See Dare v. California Department of Motor Vehicles, 191 F.3d 1167 (9th Cir. 1999). The Supreme Court's opinion in Kimel calls all of these decisions into question, and we think it best to analyze the subject afresh rather than to rehash pre-Kimel conclusions in and out of this circuit. Believing that the Supreme Court would tackle the issue before July, the second circuit declined to reconsider Muller in light of Kimel. See Kilcullen v. New York State Department of Labor, 2000 U.S. App. Lexis 2714 (2d Cir. Feb. 24, 2000). But settlements have dashed that hope; we therefore undertake independent consideration.

Whether Congress has authorized federal litigation against states is our initial question. Kimel answered yes for the ADEA, see 120 S. Ct. at 640-42, and the same answer is appropriate for the ADA. By incorporating 42 U.S.C. sec.2000e, the ADA defines persons, and thus employers, to include units of government. 42 U.S.C. sec.12111(5)(A), (7). Fitzpatrick held that sec.2000e is a sufficiently clear statement. Section 12202 adds that "[a] State shall not be immune under the eleventh amendment to the Constitution of the United States from an action in Federal or State court of competent jurisdiction for a violation of this chapter." Finally, just in case there were doubt, sec.12101(b)(4) invokes all possible sources of authority to enact the ADA, "including the power to enforce the fourteenth amendment".

On the question whether a statute such as the ADA enforces the Fourteenth Amendment, Kimel establishes two principal propositions. First, because the rational-basis test applies to age discrimination, almost all of the ADEA's requirements stand apart from the Constitution's rule. Most age discrimination is rational, and therefore constitutional, yet the Act forbids it. The ADEA therefore does not "enforce" the Fourteenth Amendment. 120 S. Ct. at 645-48. Second, there is no need for prophylactic rules to catch evasions of the rational-basis test by state governments. Congress did not find that such a problem exists, and there is no evidence of one. The ADEA therefore cannot be understood as enforcement legislation. 120 S. Ct. at 648-50. Both of these propositions are true of the ADA as well--indeed, the ADA is harder to conceive as "enforcement" of the Fourteenth Amendment than is the ADEA. Under the ADEA employers must ignore age but are free to act on the basis of attributes such as strength, mental acuity, and salary that are related to age. Hazen Paper Co. v. Biggins, 507 U.S. 604 (1993). In other words, the ADEA forbids disparate treatment but not disparate impact. EEOC v. Francis W. Parker School, 41 F.3d 1073, 1077 (7th Cir. 1994); Anderson v. Baxter Healthcare Corp., 13 F.3d 1120 (7th Cir. 1994). Likewise with the Constitution and most other employment-discrimination laws. E.g., Troupe v. May Department Stores Co., 20 F.3d 734 (7th Cir. 1994) (the Pregnancy Discrimination Act does not require accommodation). Title I of the ADA, by contrast, requires employers to consider and to accommodate disabilities, and in the process extends beyond the anti-discrimination principle. 42 U.S.C. sec.12112(b)(5)(A), (6) (defining failure to accommodate, and criteria with disparate impacts, as "discrimination"). (Some other titles of the ADA are less expansive. See Doe v. Mutual of Omaha Insurance Co., 179 F.3d 557 (7th Cir. 1999). Our concern in this case is Title I, and unelaborated references to "the ADA" are to Title I.)

A rational-basis test applies to distinctions on the ground of disability, just as to distinctions on the ground of age. Cleburne v. Cleburne Living Center, Inc., 473 U.S. 432, 439-42 (1985); Heller v. Doe, 509 U.S. 312, 319-21 (1993); United States v. Harris, 197 F.3d 870, 873-76 (7th Cir. 1999). Consideration of an employee's disabilities is proper, so far as the Constitution is concerned. See Cleburne, 473 U.S. at 444 ("governmental consideration of those differences in the vast majority of situations is not only legitimate but also desirable"). Consider this from the perspective of a university such as our defendant. A would-be professor who is not in the top 1% of the population in mental acuity is not apt to be a good teacher and scholar. Likewise it is rational for a university to favor someone with good vision over someone who requires the assistance of a reader. The sighted person can master more of the academic literature (reading is much faster than listening), improving his chance to be a productive scholar, and also is less expensive (because the university need not pay for the reader). An academic institution that prefers to use a given budget to hire a sighted scholar plus a graduate teaching assistant, rather than a blind scholar plus a reader, has complied with its constitutional obligation to avoid irrational action. But it has not complied with the ADA, which requires accommodation at any cost less than "undue hardship". 42 U.S.C. sec.12112(b) (5)(A), sec.12111(10). How the "undue hardship" defense under the ADA compares with the "bona fide occupational qualification" defense under the ADEA is an interesting question, but not one we need pursue: both statutes presumptively forbid consideration of attributes that the Constitution permits states to consider, and then (like the RFRA) require the state to carry a burden of persuasion in order to take the characteristic into account. As in Kimel, the fact that the law has made adverse action based on a characteristic "prima facie unlawful" shows the extent of its departure from the Constitution's own rule. 120 S. Ct. at 647. Like the ADEA, the ADA "prohibits very little conduct likely to be held unconstitutional," id. at 648.

The ADA's main target is an employer's rational consideration of disabilities. Rational discrimination by definition does not violate a constitutional provision that condemns only irrational distinctions based on disabilities. Congress has ample power under the Commerce Clause to forbid rational discrimination, which may bear especially heavily on a class of persons who suffer from diminished human (and often financial) capital. But to say that in devising these new rules Congress is just "enforcing" a substantive command present in sec.1 of the Fourteenth Amendment since 1868 would be a legal fiction. Boerne, Florida Prepaid, and Kimel hold that fictions do not support legislation under sec.5.

One way to distinguish the ADA from the ADEA would be to emphasize a remark in Kimel that "[o]ld age . . . does not define a discrete and insular minority because all persons, if they live out their normal life spans, will experience it." 120 S. Ct. at 645. The argument would continue that many disabilities are immutable; few people born blind acquire vision later. We do not read the Court's observation in Kimel as distinguishing among characteristics that are subject to rational-basis review; instead the Court offered the observation as one reason why earlier cases had applied the rational-basis test to age. Because Cleburne held that the rational-basis test likewise governs disabilities, the reasoning behind that opinion need not come back into consideration. We know from Cleburne that rational distinctions based on disabilities comport with the Constitution. What is more, many disabilities come and go, or progress with time. Beethoven did not become deaf, or Milton blind, until middle age. Erickson's medical problem affected her for a number of years but not for a lifetime (if only because medical treatment may have succeeded, or because after menopause it would have lost significance). One can imagine an argument under sec.5 for a federal law dealing with discrimination against persons with life- long disabilities, but the ADA is not such a law-- not only because it extends beyond permanently disabled persons, but also because "discrimination" as the ADA defines it, see sec.12112(b), has little in common with "discrimination" in constitutional law.

To see this, consider the role of intent. When a state law or practice does not expressly concern a particular characteristic (such as race, sex, age, or disability), but has a disparate impact on persons with that characteristic, the plaintiff in constitutional litigation must establish that the state intends to discriminate on the basis of that characteristic. See, e.g., Personnel Administrator v. Feeney, 442 U.S. 256 (1979) (sex); Washington v. Davis, 426 U.S. 229 (1976) (race). Things are otherwise under the ADA, which not only demands accommodation (which forces the employer to consider, rather than ignore, disabilities) but also prohibits any rule or practice that has a disparate impact, unless the rule is "job-related for the position in question and is consistent with business necessity". 42 U.S.C. sec.12112(b)(6). See Washington v. Indiana High School Athletic Ass'n, 181 F.3d 840 (7th Cir. 1999) (under the ADA the plaintiff need not show that the governmental body intended to discriminate on account of disability). Cases such as Feeney and Davis hold that the Equal Protection Clause does not forbid laws and practices that have a disparate impact; but the ADA does forbid them.

By requiring that employers accommodate rather than disregard disabilities, the ADA is a cousin to the RFRA. Smith held that demands for accommodation and claims of disparate impact have no constitutional footing under the Free Exercise Clause; it takes express or intentional discrimination to violate that provision. See also Church of the Lukumi Babalu Aye, Inc. v. Hialeah, 508 U.S. 520 (1993). Congress then enacted the RFRA, which requires every unit of government to justify any law or practice that burdens a person's exercise of religion, "even if the burden results from a rule of general applicability". 42 U.S.C. sec.2000bb-1(a). This requires a state to accommodate religiously motivated behavior unless it can show a "compelling" reason for neutrality between religious and secular conduct. Boerne responded that Congress may not redefine the constitutional rule under the rubric of "enforcement." What the RFRA did for religion, the ADA does for disabilities. In neither situation does the Constitution forbid neutral laws or practices that create disparate impacts; in neither situation does the Constitution require accommodation. Both the RFRA and the ADA replace the Constitution's approach with a prohibition of disparate impact and jettison neutrality in favor of accommodation. The RFRA's demand for a "compelling governmental interest", 42 U.S.C. sec.2000bb-1(b)(1), made it harder for a government to prevail than do the ADA's requirements (job-relatedness, business necessity, and undue hardship), but there is a countervailing difference that makes the ADA the more adventuresome. The Free Exercise Clause forbids all intentional discrimination against religious practices; the Equal Protection Clause has no similar rule about disabilities. Rational discrimination against persons with disabilities is constitutionally permissible in a way that rational discrimination against religious practices is not. This makes the ADA harder than the RFRA to justify under sec.5, for "[i]t is precisely in a close case that the independent judgment of Congress on a constitutional question should make a difference." Michael W. McConnell, Institutions and Interpretation: A Critique of City of Boerne v. Flores, 111 Harv. L. Rev. 153, 155 (1997). See also Stephen L. Carter, The Morgan "Power" and the Forced Reconsideration of Constitutional Decisions, 53 U. Chi. L. Rev. 819 (1986). Some of the Justices and several careful scholars believe that the rule of decision in the RFRA is the Constitution's own. See Boerne, 521 U.S. at 544 (O'Connor, J., dissenting), 565 (Souter, J., dissenting). Others who support the majority position in Smith acknowledge that the question is difficult. See generally Symposium, Reflections on City of Boerne v. Flores, 39 William & Mary L. Rev. 597 (1998). But no one believes that the Equal Protection Clause establishes the disparate-impact and mandatory- accommodation rules found in the ADA. The statute is outside the boundaries of constitutional discourse in a way that the RFRA was not. If the RFRA and the ADEA exceed the sec.5 power, then so does the ADA--at least to the extent it extends beyond remedies for irrational discrimination.

Well, then, can the ADA be sustained as reasonable prophylactic legislation? Because the ADA requires accommodation, forbids practices with disparate impact, and disregards the employer's intent, it is harder than the ADEA to characterize as a remedial measure. The ADEA was a real anti- discrimination law; unless age was held against the employee, there was no violation. The ADA goes beyond the anti-discrimination principle, a step that requires reason to think that only by going to these lengths is it possible to implement the core constitutional rule. Yet just as for the ADEA, Congress did not find that states have adopted clever devices that conceal irrational discrimination. The legislative findings in 42 U.S.C. sec.12101 contain not a word about state governments. Congress did find that persons with disabilities have been discriminated against; it found the same in the ADEA for age. What it did not find is that the practices labeled "discrimination" are irrational (as that term works under the Equal Protection Clause) or that states are major offenders--a critical inquiry not only under Kimel but also under Florida Prepaid. Instead, Congress used the word "discrimination" in sec.12101, and Committees of Congress used that word in the legislative history, to refer to any disadvantage that accompanies a disability. For example, the statement in H.R. Rep. No. 101-485(II), 101st Cong. 2d Sess. 37 (1990), that "inconsistent treatment of people with disabilities by different State or local government agencies is both inequitable and illogical for a society committed to full access for people with disabilities" means only that different public bodies treated persons differently, because the Rehabilitation Act applied to some persons but not others; it does not mean that either treatment was unconstitutional. "Inconsistent" is not a synonym for irrational--especially not when it was a federal statute that induced the inconsistency on which the Committee remarked.

Just as in Kimel, legislative statements about discrimination consist "almost entirely of isolated sentences clipped from floor debates and legislative reports." 120 S. Ct. at 649. These snippets use the word "discrimination" in a way that fails to distinguish between rational distinctions (which the Constitution allows) and irrational ones (which it forbids). The sort of findings that would permit adoption of the ADA as a precautionary measure, after the fashion of the Voting Rights Act, see South Carolina v. Katzenbach, 383 U.S. 301 (1966), must establish that states have been able to disguise forbidden discrimination as the permissible kind. Nothing in the legislative findings, or the debates preceding the ADA's adoption, shows (or even asserts) that state governments engaged in deception that prevented victims of irrational discrimination from obtaining a remedy. Findings underlying Title VII were more substantial, and because employers frequently disguised their resort to racial criteria it is easier to justify the disparate-impact features of Title VII as remedial measures. In re Employment Discrimination Litigation, 198 F.3d 1305 (11th Cir. 1999), concludes accordingly that sec.5 supports the disparate-impact rules under Title VII, as well as the disparate-treatment rules addressed in Fitzpatrick v. Bitzer. We leave that question for another day and hold only that the background of the ADA does not meet the standards that Boerne and Kimel set for using sec.5 to enact prophylactic legislation.

From all of this it follows that the ADA does not "enforce" the Fourteenth Amendment, and from Seminole Tribe it follows that the Eleventh Amendment and associated principles of sovereign immunity block private litigation against states in federal court. But Northeastern Illinois University must understand the limits of this holding. The ADA is valid legislation, which both private and public actors must follow. Even if the Supreme Court should overrule Garcia v. San Antonio Metropolitan Transit Authority, 469 U.S. 528 (1985), and return to the view of National League of Cities v. Usery, 426 U.S. 833, 852 (1976), that laws resting only on the Commerce Clause cannot "directly displace the States' freedom to structure integral operations in areas of traditional governmental functions," the University still would be bound by the ADA, for running a university is no more a core governmental function than is running a railroad. See United Transportation Union v. Long Island R.R., 455 U.S. 678 (1982). Like most railroads, most universities in the United States are private. All our holding means is that private litigation to enforce the ADA may not proceed in federal court. Erickson may repair to Illinois court--for although states may implement a blanket rule of sovereign immunity, see Alden v. Maine, 527 U.S. 706 (1999), Illinois has not done this. Having opened its courts to claims based on state law, including its own prohibition of disability discrimination by units of state government, see 775 ILCS 5/1-102, 5/2- 101(B)(1)(c), Illinois may not exclude claims based on federal law. Howlett v. Rose, 496 U.S. 356, 367-75 (1990); FERC v. Mississippi, 456 U.S. 742, 759-69 (1982); Testa v. Katt, 330 U.S. 386 (1947). Moreover, the United States may enforce the ADA against the University and other state actors through federal litigation. West Virginia v. United States, 479 U.S. 305, 311 n.4 (1987). But Erickson has not enlisted the United States as her champion (its intervention was for the purpose of defending Erickson's right to sue in her own name), so this suit belongs in state court.

Reversed Diane P. Wood, Circuit Judge, dissenting.

The Americans with Disabilities Act, or ADA, 42 U.S.C. sec. 12111 et seq., stands at the intersection of two lines of cases that address Congress's power under section 5 of the Fourteenth Amendment to abrogate the Eleventh Amendment immunity of the states. Laws that fall within the section 5 power may abrogate the States' Eleventh Amendment immunity from suit, if Congress has made its intent to abrogate "unmistakably clear" in the language of the statute. See City of Boerne v. Flores, 521 U.S. 507 (1997); Atascadero State Hospital v. Scanlon, 473 U.S. 234, 242 (1985). When the question has been whether Title VII of the Civil Rights Act represents a valid use of Congress's power under section 5, courts have answered in the affirmative. See, e.g., Fitzpatrick v. Bitzer, 427 U.S. 445, 456-57 (1976); In re Employment Discrimination Litigation Against State of Alabama, 198 F.3d 1305, 1324 (11th Cir. 1999) (finding that disparate impact analysis is a valid prophylactic measure and thus that this aspect of Title VII, equally with the disparate treatment branch, is a valid exercise of section 5 power).

On the other hand, the Supreme Court has recently ruled that the Age Discrimination in Employment Act, or ADEA, 29 U.S.C. sec. 621-34, exceeded Congress's section 5 powers and thus could not as a matter of law override the State's Eleventh Amendment immunity. Kimel v. Florida Board of Regents, 120 S. Ct. 631 (2000). The question before us today, as the majority recognizes, is which line of authority to apply to yet another statute, the ADA. This is plainly a delicate and difficult issue, as the Supreme Court itself appeared to have signaled when it granted certiorari in Florida Dept. of Corrections v. Dickson, 120 S. Ct. 976 (2000), and in Alsbrook v. Arkansas, 120 S. Ct. 1003 (2000). We must therefore decide this case without the prospect of immediate guidance from Washington. For the reasons I explain below, I conclude that Title I of the ADA falls within Congress's section 5 powers under the principles the Court has articulated. I would therefore find that Erickson is entitled to bring her ADA suit against Northeastern Illinois University consistently with the Eleventh Amendment, and I respectfully dissent.

* Although the literal language of the Eleventh Amendment addresses only the question of the extent of the judicial power of the United States (which "shall not be construed to extend to any suit in law or equity, commenced or prosecuted against one of the United States by Citizens of another State, or by Citizens or Subjects of any Foreign State," U.S. Const. amend. XI), the Supreme Court has held in a recent line of decisions that the meaning of this part of the Constitution is not limited to the precise words of the text. Instead, the Eleventh Amendment reflects the structural fact that each state is a sovereign entity within the federal system, and as such, each state enjoys sovereign immunity from suit except insofar as its immunity has legitimately been curtailed. See Seminole Tribe v. Florida, 517 U.S. 44, 54 (1996); Alden v. Maine, 119 S. Ct. 2240, 2253-54 (1999); Florida Prepaid Postsecondary Education Expense Board v. College Savings Bank, 119 S. Ct. 2199, 2204 (1999).

There are a number of ways in which sovereign immunity can be overcome consistently with the law: the state might consent to suit; to much the same effect, it might choose to waive its sovereign immunity; or Congress might enact legislation that abrogates the state's immunity.1 Only the last of those options is relevant here. Abrogation is constitutionally possible only in narrow circumstances. First, Congress must make its intent to abrogate "unmistakably clear" in the language of the statute. See Kimel, 120 S. Ct. at 640 (citing Dellmuth v. Muth, 491 U.S. 223, 228 (1989), and quoting from Atascadero, 473 U.S. at 242). Second, it must act pursuant to a valid grant of constitutional power. Kimel, 120 S. Ct. at 642; City of Boerne, 521 U.S. 507, 519; Green v. Mansour, 474 U.S. 64, 68 (1985). Here, everyone agrees that the only source of congressional power at issue is section 5 of the Fourteenth Amendment. Cf. Florida Prepaid, 119 S. Ct. at 2205.

In Kimel, the Court found that the ADEA satisfied the "clear statement" requirement for abrogation. 120 S. Ct. at 640-42. The majority finds, and I agree, that the same is true of the ADA. Unlike the majority, however, I also conclude that Congress legitimately used its power under section 5 of the Fourteenth Amendment when it made the ADA applicable to the states.

As I have already noted, we know that Title VII represents a valid exercise of Congress's section 5 power to abrogate the Eleventh Amendment immunity of the states, but the ADEA does not. The Kimel Court made the latter finding because, following City of Boerne, it concluded that the ADEA was a measure that went beyond either enforcement of the Fourteenth Amendment or valid prophylactic measures designed to prevent violations of the Constitution. See Kimel, 120 S. Ct. at 645, 648-49. In Florida Prepaid, the Court explained the difference between valid efforts to exercise section 5 powers and those that go beyond the constitutional limits as follows:

While the line between measures that remedy or prevent unconstitutional actions and measures that make a substantive change in the governing law is not easy to discern, and Congress must have wide latitude in determining where it lies, the distinction exists and must be observed. There must be a congruence and proportionality between the injury to be prevented or remedied and the means adopted to that end.

119 S.Ct. at 2205 (quoting from City of Boerne, 521 U.S. at 519-20).

While the majority appears to concede that Kimel should guide our decision with respect to the ADA, its reading of Kimel overlooks important qualifications on that decision. The majority sees Kimel as a case holding that virtually all discrimination that is subject to rational basis review for equal protection clause purposes is outside the scope of Congress's section 5 powers. Ante, at 948. I find no hint of this in Kimel; to the contrary, after recognizing that age discrimination is subject to rational basis review, the Court took pains to analyze the ADEA in detail before finding that it cannot be sustained against the states as a valid exercise of the section 5 powers. That analysis would have been entirely beside the point if the mere fact of rational basis review was enough to decide the case. Furthermore, the majority here, in rejecting the idea that the accommodation provisions of the ADA could be sustained under section 5 (ante at 950) ignores the express holding of Kimel that "we have never held that section 5 precludes Congress from enacting reasonably prophylactic legislation." 120 S. Ct. at 648. Last, the majority appears to hold that virtually all antidiscrimination statutes that focus on disparate impact, rather than intentional disparate treatment, exceed Congress's section 5 powers. In so doing, it has created a square conflict with the Eleventh Circuit's decision in Employment Discrimination, supra, 198 F.3d at 1324.

Kimel provides the analytical approach for assessing whether a statute addressing discrimination is a valid exercise of the section 5 power. Looking at both the legislative record and the language of the pertinent statute, the Kimel Court first asked whether the substantive requirements of the statute were proportionate to any unconstitutional conduct that the statute could have targeted. 120 S. Ct. at 645. It looked to earlier decisions that had considered the constitutional implications of age discrimination and found it significant that all had upheld age distinctions against constitutional challenges. See Gregory v. Ashcroft, 501 U.S. 452 (1991); Vance v. Bradley, 440 U.S. 93 (1979); Massachusetts Bd. of Retirement v. Murgia, 427 U.S. 307 (1976) (per curiam). Second, it consulted the legislative record to see if it revealed either (1) a pattern of age discrimination committed by the states or (2) "any discrimination whatsoever that rose to the level of constitutional violation." 120 S. Ct. at 648-50. Finding neither element present, the Court concluded that Congress did not in the ADEA validly abrogate the states' sovereign immunity.

Following this roadmap, one can see that the ADA differs critically from the ADEA in the areas the Supreme Court deemed significant. The first question concerns the level of constitutional protection the Supreme Court has recognized in prior cases for persons with disabilities. With that standard in mind, the next question is whether the ADA represents a proportionate response to the likelihood of constitutional violations.

The leading case on the equal protection dimensions of disability discrimination is City of Cleburne v. Cleburne Living Center, Inc., 473 U.S. 432 (1985). Although, as the majority observes, the Court ultimately decided that rational basis review was proper for the ordinance in that case, the majority finds the Court's reasoning to be irrelevant, ante at 949. The majority also pays no heed to the fact that the Court struck down the Cleburne ordinance because it unconstitutionally discriminated against the mentally retarded (clearly illustrating that legislation prohibiting discrimination with respect to a category that receives rational basis review might indeed be enforcing the Constitution). I cannot dismiss either aspect of Cleburne so readily.

The specific question before the Court in Cleburne was whether a local ordinance that required a special use permit for a home for the mentally retarded, but that imposed no such requirement for many similar uses, violated the equal protection rights of the mentally disabled. The Court held that mental retardation should not be treated as a "quasi-suspect classification" for equal protection purposes, but it nevertheless found that the ordinance failed rational basis scrutiny, because the permit requirement "rest[ed] on an irrational prejudice against the mentally retarded . . . ." Cleburne, 473 U.S. at 450, 105 S.Ct.3249.2 In coming to that conclusion, the Court subjected the city's proffered reasons in defense of the ordinance to careful scrutiny, even while it avoided introducing undue rigidity into its analysis by using terms like "suspect" or "quasi-suspect" classifications--terms which the Court later pointed out had sometimes given rise to the erroneous notion that scrutiny that was strict in theory was often fatal in fact. See Adarand Constructors, Inc. v. Pena, 515 U.S. 200, 237 (1995).

Both the rationale of Cleburne and the nature of disability discrimination itself, as outlined in the congressional findings and legislative history of the ADA, highlight important differences between disability and age as bases for differential treatment, and they reveal, contrary to the majority's surprising suggestion, that the ADA is indeed a statute designed to prohibit irrational discrimination.

As the Kimel Court observed, older persons "have not been subjected to a history of purposeful unequal treatment." 120 S. Ct. at 645 (citing Murgia, 427 U.S. at 313, quoting San Antonio Independent School Dist. v. Rodriguez, 411 U.S. 1, 28 (1973)). In contrast, Congress found in the ADA that disabled persons have been "subjected to a history of purposeful unequal treatment," "in such critical areas as employment, housing, public accommodations, education, transportation, communication, recreation, institutionalization, health services, voting, and access to public services." 42 U.S.C. sec. 12101. Second, harking back to the well known idea in United States v. Carolene Products, 304 U.S. 144, 152-53 n.4 (1938), in no meaningful sense of the term can the elderly be regarded as a "discrete and insular minorit[y]"; to the contrary, as Kimel notes, "all persons, if they live out their normal life spans, will experience [old age]." 120 S. Ct. at 645. This is a strong reason to believe that the normal political processes are adequate to protect the interests of the elderly and that they will not be singled out for unconstitutionally discriminatory treatment.

The disabled stand in a distinctly different position. Not everyone is or will become disabled. And the fact that some disabilities arise later in life and some do not persist for a lifetime does not make them the equivalent of the inexorable aging process. The point is that Congress found that those who are disabled will suffer during the time they are disabled from the same invidious discrimination that has haunted racial minorities and women. The ADA reflects Congress's finding that society has the ability to, and has historically, "tended to isolate and segregate individuals with disabilities." 42 U.S.C. sec. 12101.

There are other reasons as well to conclude that the ADA is a permissible exercise of Congress's section 5 power. Apart from the salient differences between age and disability as bases for categorization, the two statutes fare quite differently under the proportionality analysis required by Boerne and Kimel. The broad sweep of the ADEA caused the Supreme Court to find that it was not a proportional response to the problem of age discrimination. The ADEA prohibits all employment discrimination on the basis of age against persons in the protected class (those above the age of 40). 29 U.S.C. sec. 623(a)(1). The only tempering of this rule appears in the statutory rules allowing an employer to justify age-based distinctions if it shows either a substantial basis for believing that all or nearly all employees above a given age lack the qualifications required for the position or that reliance on the age classification is necessary because individual testing for qualifications is highly impractical. Kimel, 120 S. Ct. at 647 (citing Western Air Lines v. Criswell, 472 U.S. 400, 422 (1985)). The EEOC's implementing regulations, as well as cases decided under the ADEA, make it clear that these exceptions were intended to be narrow ones. See 29 C.F.R. sec. 1625.6(a); see also Western Air Lines, 472 U.S. at 422.

The ADA adopts a more nuanced approach to the problem of disability discrimination. An employer is entitled to treat a disabled person differently--indeed, even to deny employment to the person on that basis--if there are no reasonable accommodations that will permit the individual to do the job and she cannot handle the job without accommodations. 42 U.S.C. sec. 12113. See, e.g., Stewart v. County of Brown, 86 F.3d 107, 112 (7th Cir. 1996); Pond v. Michelin North America, Inc., 183 F.3d 592, 596 (7th Cir. 1999); Sieberns v. Wal-Mart Stores, Inc., 125 F.3d 1019, 1022 (7th Cir. 1997). Thus, while an employer discriminating on the basis of age must demonstrate that it would be "highly impractical" not to do so, an employer making distinctions on the basis of disability need only show that "reasonable steps" of accommodation, such as modifying work schedules, training materials, facilities, or policies, will not work. See 42 U.S.C. sec.sec. 12113, 12111. The incorporation of a reasonableness standard in the duty to accommodate, which itself modifies the duty not to discriminate on the basis of disability, is essentially a legislative incorporation of the proportionality test required under the Constitution. It also illustrates, contrary to the majority's suggestion, that the duty to accommodate is not a command to give "special" treatment; instead, it spells out the way that discrimination is to be avoided. I would therefore find that the ADA meets the first part of the Kimel analysis.

The second question under Kimel requires us to consider whether the legislative record reveals either a pattern of age discrimination committed by the states or "any discrimination whatsoever that [rises] to the level of constitutional violation." 120 S. Ct. at 649. Here, although the evidence is stronger on the second point than the first, the record shows both kinds of disability discrimination.

With respect to the first question (i.e. legislative findings pertaining specifically to state behavior), the legislative record is admittedly sparse. Nevertheless, the House Report notes that "inconsistent treatment of people with disabilities by different state or local government agencies is both inequitable and illogical." H.R. Rep. No. 101-485 (II). More importantly, the express congressional findings with respect to pervasive discrimination address many areas that are controlled to a significant degree by state and local governments. For example, Congress identified discrimination in education as a particular problem. See 42 U.S.C. sec. 12101(3). Education in this country is overwhelmingly an enterprise of state and local government.3 Another sector singled out in the statute was health services, see 42 U.S.C. sec. 12101(3), in which state and local governments also play a powerful role.4 The story is similar for transportation, which is also mentioned in sec. 12101(3).5 Congress's specific attention to sectors with such a substantial state and local governmental presence indicates that it knew that government action at the state level was an important part of the problem it was addressing.

The other evidence the Kimel Court found lacking for the ADEA--a record of discrimination that reveals constitutional violations--is present in abundance for the ADA. It would be hard to imagine greater scrutiny than Congress gave to the harm caused by disability discrimination when it passed the ADA. Its findings explain in painstaking detail the extent of the evil. See 42 U.S.C. sec. 12101.6 We give congressional findings substantial deference, because Congress "is far better equipped than the judiciary to amass and evaluate the vast amounts of data bearing upon legislative questions." Turner Broadcasting Systems v. F.C.C., 520 U.S. 180, 195 (1997). This is the legislative task the Supreme Court contemplated in Cleburne, where it held that the way disabled people are "to be treated under the law is a difficult and often a technical matter, very much a task for legislators guided by qualified professionals and not by the perhaps ill-informed opinions of the judiciary." Cleburne, 473 U.S. at 442-43.

The ADA's legislative findings distinguish the ADA from both the ADEA and RFRA, the statute before the Court in City of Boerne. Like the ADEA and unlike the ADA, Congress did not make findings in the RFRA about the seriousness or scope of discrimination against religious persons. See 42 U.S.C. sec.sec. 2000bb to 2000bb-4. As I have already noted, in the ADEA Congress never identified "any discrimination whatsoever that rose to the level of constitutional violation." Kimel, 120 S. Ct. at 649. The only evidence the Kimel Court found showing the harm at which the ADEA was aimed was a few "isolated sentences clipped from floor debates and legislative reports." Id. When formulating the ADA, in contrast, Congress compiled an immense legislative record. It examined all this evidence and found that "[t]he severity and pervasiveness of discrimination against people with disabilities [was] well documented." H.R. 101-485 (II). This factor therefore points toward a conclusion that the legislative basis for a valid exercise of Congress's section 5 powers is present for the ADA, even though it was not for the ADEA or RFRA.

Before leaving this subject, it is important to note that the majority has elevated a single point in the legislative history to dispositive significance: the absence of a statement somewhere to the effect that "we are passing this law because we need to correct discrimination on the basis of disability committed by the states." I see nothing in Kimel that gives such primacy to this single point. Combining the explicit coverage of sectors in which the states are the principal actors, with the deliberate decision of Congress to make the states subject to the statute, and finally with the enormous legislative record documenting the depth of the problem of disability discrimination, I find the second part of the Kimel approach to be satisfied for the ADA.

Given its conclusion about the Eleventh Amendment, the majority does not reach the last question that was presented in this case, which was whether the analysis that applies to an Eleventh Amendment argument directed at the general prohibition in the ADA against discrimination is different from the analysis appropriate to the accommodation provisions of the Act. Because I would reject the general Eleventh Amendment defense, I add a brief word on this point. In my view, because the accommodation duty and the duty to avoid discrimination are nothing more than two sides of the same coin, the answer is no.

The ADA defines discrimination to include "not making reasonable accommodations to the known physical or mental limitations of an otherwise qualified individual with a disability who is an applicant or employee, unless . . . [the] covered entity can demonstrate that the accommodation would impose an undue hardship on the operation of the business of such covered entity." 42 U.S.C. sec. 12112(b)(5)(A). The Act also provides that an employer may defend against a charge of discrimination by showing that its goals require discrimination--that they "cannot be accomplished by reasonable accommodation." 42 U.S.C. sec. 12113(a).

The University argues that this statutory accommodation process is unconstitutional under Printz v. United States, 521 U.S. 898 (1997), because it violates the Tenth Amendment by forcing state officials to administer a federal regulatory scheme. In my view, however, the Printz model has no bearing on the question before us. The flaws the Court identified in Printz included the act of conscripting state officials to administer a federal program, the effective reallocation of duties from the branches of the federal government to which the Constitution assigned them to the state officials, and the conferral of policy-making authority on the state officials without adequate guidance. The Printz Court found that forcing the state to implement this type of regulatory system violated the principles of separation of powers and dual sovereignty. Id. at 922, 932, 930, 117 S.Ct.2365.

The ADA does not establish anything like the regulatory scheme for handguns at issue in Printz. The ADA is instead a straightforward law prohibiting discrimination on the part of all employers, private and governmental alike, and defining the way the prohibition must be implemented. It provides the employers with precise definitions to follow: a reasonable accommodation is one tailored to the discrimination issue before the employer, which does not "impose an undue hardship on the operation [of the employer's business]." 42 U.S.C. sec. 12112(b)(5)(A). Unlike the regulatory system before the Printz Court, the ADA does not confer any special powers on employers in general or on state employers in particular. Employers are not administering a federal benefit by providing a reasonable accommodation; they are refraining from discrimination and to some degree taking preventative measures. There is no duty to accommodate that is separate from the general obligation to avoid discrimination against the disabled.

It bears repeating that, for this purpose, state employers stand in exactly the same position as private employers. As this court held in Travis v. Reno, 163 F.3d 1000, 1004-05 (7th Cir. 1998), federal law may pervasively regulate states as market participants; the anti- commandeering law of Printz only comes into play when the federal government calls on the states to use their sovereign powers to implement a federal regulatory program. In Travis, which came to the result later endorsed by the Supreme Court in Reno v. Condon, supra, we concluded that the Drivers Privacy Protection Act (DPPA) did not violate the Tenth Amendment. The DPPA requires disclosure of certain records by the state, and so necessarily forces the state to come up with a system of determining which records should be disclosed, as well as how best to disclose them. The system was found constitutional because it affects states in their role as owners of databases, not in their role as governments. Condon, 120 S. Ct. at 672; Travis, 163 F.3d at 1004.

Though the ADA forces the states to comply with a federal regulation, it affects the states in their role as employers, not in their role as governments. Federal regulations of states acting as employers have been upheld in the past. In Garcia v. San Antonio Metropolitan Transit Authority, 469 U.S. 528 (1985), the Court held that state employers may be forced to follow the federal Fair Labor Standards Act's wage and hour rules. Nothing in the recent line of Eleventh Amendment decisions undermines that rule. To the contrary, in Alden v. Maine the Court went out of its way to reaffirm that "[t]he constitutional privilege of a State to assert its sovereign immunity in its own courts does not confer upon the State a concomitant right to disregard the Constitution or valid federal law." 119 S. Ct. at 2266. Instead, the Court assumed that the states would ordinarily live up to their duties under federal law as a matter of good faith, and it noted that enforcement of federal obligations by the federal government remains permissible under the constitutional design. Id. at 2267. The fact of dual sovereignty does not, therefore, carry with it any implication that states are allowed to disregard or to frustrate valid federal programs. See City of New York v. United States, 179 F.3d 29, 35 (2d Cir. 1999).

By defining discrimination in part as not making reasonable accommodations to disabled employees, the ADA does impose costs on employers, including the states. Employers must affirmatively act to alter any practices they have in place that discriminate against the disabled. Of course, this makes a great deal of sense. Just because an employer has a discriminatory practice, such as maintaining steep stairways or only offering breaks at wide intervals and therefore not allowing diabetics to take their medication, does not mean that the employer should be able to continue such a discriminatory practice without violating the ADA, any more than an employer's refusal in the past to construct a women's restroom would justify a refusal to hire female employees. The ADA allows an employer to adjust the workplace environment on a case-by-case basis, adopting only those changes that are reasonably necessary to refrain from discriminating against the disabled individual or individuals in question.

The ADA hardly broke new ground when it incorporated this type of affirmative duty. The Equal Protection Clause often requires states to take affirmative measures to eliminate or prevent discriminatory systems. For example, states with racially discriminatory reapportionment plans must redraw their congressional districts. See, e.g., Shaw v. Reno, 509 U.S. 630, 652 (1993) (holding that the state's reapportionment plan might violate the Equal Protection Clause). The logic of the University's argument here would, if taken to its limits, call into question every affirmative injunction a court has ever entered to prevent threatened future violations of the constitutional guarantee of equal protection of the laws. Nothing in the Supreme Court decisions on which the University relies even hints at such a radical result. Similarly, the First Amendment guarantee of the right of free exercise of religion carries with it an implied duty on the part of the state to make reasonable adjustments. See, e.g., Sherbert v. Verner, 374 U.S. 398, 403- 04 (1963); Wisconsin v. Yoder, 406 U.S. 205, 231 (1972); Church of the Lukumi Babalu Aye v. City of Hialeah, 508 U.S. 520, 546 (1993); Zorach v. Clauson, 343 U.S. 306, 313-14 (1952). Boerne does not overrule these direct constitutional rulings.

Last, as I indicated above, I do not read any of the Supreme Court's recent decisions as overruling prior rulings that have upheld congressional legislation prohibiting measures with a discriminatory impact as valid exercises of the section 5 power. As the Eleventh Circuit explained in Employment Discrimination, "disparate impact analysis was designed as a 'prophylactic' measure." 198 F.3d at 1321 (citing Connecticut v. Teal, 457 U.S. 440, 449 (1982), Albemarle Paper Co. v. Moody, 422 U.S. 405, 417 (1975), and Griggs v. Duke Power Co., 401 U.S. 424, 435 (1971)). The Eleventh Circuit went on to explain that even though, in a disparate impact case, "the plaintiff is never explicitly required to demonstrate discriminatory motive, a genuine finding of disparate impact can be highly probative of the employer's motive since a racial 'imbalance is often a telltale sign of purposeful discrimination.'" Id. (citing International Brotherhood of Teamsters v. United States, 431 U.S. 324, 339-40 n.20 (1977)). It found from this that the disparate impact provisions of Title VII are preventive rules that have the necessary congruence between the means used and the constitutional violation to be addressed (intentional discrimination). Id. at 1322. Nothing in Kimel comes close to suggesting that the Court was overruling this long line of its own authority, upon which the Eleventh Circuit carefully relied, and I am not prepared to take that step in the present case.

For all these reasons, I therefore respectfully dissent from the majority's conclusion that the Eleventh Amendment bars Erickson's suit against Northeastern University.