BREYER, J., dissenting

the significance of the numerical results. See App. 93-94; Gingles, 478 U. S., at 57, nn. 25 and 26.

Regardless, as the majority says, the District Court found the statistical evidence inconclusive and "conflicting." 922 F. Supp., at 1567. And the District Court conceded the existence of "some degree of vote polarization." Ibid. (It simply said that the "degree” was not “alarming."" Ibid.) That African-American incumbents were reelected does not, without more, disprove polarization. Gingles, supra, at 75 ("[T]he election of a few minority candidates does not "necessarily foreclose the possibility of dilution of the black vote...""") (quoting S. Rep. No. 97-417, p. 29, n. 115 (1982), in turn quoting Zimmer v. McKeithen, 485 F. 2d 1297, 1307 (CA5 1973) (en banc), aff'd sub nom. East Carroll Parish School Bd. v. Marshall, 424 U. S. 636 (1976) (per curiam)); 478 U. S., at 75 (citing S. Rep. No. 97-417, supra, at 29, n. 115) (listing incumbency as a special factor in assessing vote polarization).

The majority says that, despite this evidence, the District Court's findings-of no §2 violation and no §5 violation—are adequately supported. Ante, at 94, 97. But that is because the District Court asked the wrong question. We need not decide whether the evidence shows the failure to create a second majority-minority district violates §2. Cf. ante, at 90-95. (Nor, for that matter, need we decide whether the consequent reduction of such districts from 1 in 10 to 1 in 11 would, other things being equal, violate §5—which it might do. Cf. ante, at 95-98.) The question is not about whether the evidence proves §2 in fact requires two majorityminority districts. The question is whether the evidence is strong enough to justify a legislature's reasonable belief that that was so. The record rather clearly demonstrates a "strong basis in the evidence" for believing that §2 or §5 required two majority-minority districts. The legislature thus could very reasonably have believed that was so. And, that is what I had believed the law, as set forth in this

BREYER, J., dissenting

Court's opinions, required as legal justification for a district that otherwise would violate the basic predominant factor test of Miller.

This legal distinction-between whether a plan really violates § 2 or might well violate § 2-may seem technical. But it is not. A legal rule that permits legislatures to take account of race only when §2 really requires them to do so is a rule that shifts the power to redistrict from legislatures to federal courts (for only the latter can say what §2 really requires). A rule that rests upon a reasonable view of the evidence (i. e., that permits the legislature to use race if it has a "strong basis" for believing it necessary to do so) is a rule that leaves at least a modicum of discretionary (racerelated) redistricting authority in the hands of legislators. Again (and at a minimum), the District Court's use of the wrong test requires vacating its judgment.

C

To create a second majority-minority district is not impractical nor would doing so significantly interfere with other important districting objectives. The easiest way to understand why this is so is to look at three plans that I have placed in the Appendix, infra. I shall call the Georgia Legislature's 1991 two-district reapportionment Plan A. Appendix, 1991 Plan, infra. I shall call the one-district plan adopted by the court Plan B. Appendix, 1995 Court Plan, infra. And I shall call the two-district Illustrative Plan proposed by the Justice Department Plan C. Appendix, Illustrative Plan, infra. Inspection of the three plans suggests that the District Court's plan (B) is very similar to the other two (A and C) but for one critical feature, namely, that it has one majority-minority district rather than two.

Now consider the three plans in respect to each of the five districting considerations that the District Court called traditional and important. They are: (a) retaining one district in each corner of the State; (b) creating an urban minor

BREYER, J., dissenting

ity district; (c) maintaining political subdivisions; (d) protecting incumbents; and (e) maintaining traditional district cores. 922 F. Supp., at 1564-1565.

All three plans are identical in respect to the first two considerations. Each maintains districts in three of the four state corners; each creates at least one urban minority district. Plan B-the District Court's plan-is marginally superior in respect to the third criterion (maintaining political subdivisions). Plan B splits six counties within the Atlanta area but none outside the Atlanta area. Id., at 1564. Plan C splits two counties (Bibb and Muscogee) outside the Atlanta area. (Appellants, however, advance nonracial justifications for the latter splits.)

Plan C is superior to Plan B in respect to the remaining two considerations. Plan C displaces no incumbents. Plan B displaces three incumbents (including two AfricanAmericans). Plan C maintains all district cores. Plan B moves many more Georgians into new districts.

Plan C has certain other advantages: It maintains, as provided in the legislature's 1991 plan, 138 of Georgia's 159 counties. Plan B maintains 123. Plan C has greater population uniformity among its districts. And, of course, Plan C provides for two majority-minority districts-the number the legislature provided in two of its three redistricting plans.

I add one point. This is not a suit in which there are claims of interference with the right to cast a ballot or "dilution" of the majority's vote. Cf. White v. Regester, 412 U. S. 755 (1973); Reynolds v. Sims, 377 U. S. 533 (1964); and Gomillion v. Lightfoot, 364 U. S. 339 (1960); see also Karlan & Levinson, 84 Calif. L. Rev., at 1212-1216. Rather, the legislature's plans, insofar as they were race conscious, sought only to prevent what the legislature could reasonably have believed to be unlawful vote dilution-i. e., to prevent a violation of VRA §2, or perhaps §5. See Tr. 103 (Oct. 30, 1995) (testimony of Rep. Sanford Bishop). Given this fact and given the three sets of considerations just mentioned, I do

BREYER, J., dissenting

not see how the majority, consistently with Upham, can affirm the District Court's determination.

II

In other cases dissenting judges have expressed concerns that the Court's holdings and particularly its test—"predominant racial motive"-would prove unworkable, that they would improperly shift redistricting authority from legislatures to courts, and that they would prevent the legitimate use (among others the remedial use) of race as a political factor in redistricting, sometimes making unfair distinctions between racial minorities and others. See, e. g., Shaw v. Reno, 509 U. S., at 676–679 (STEVENS, J., dissenting); id., at 679-687 (SOUTER, J., dissenting); Miller, 515 U. S., at 929 (STEVENS, J., dissenting); id., at 934 (GINSBURG, J., dissenting); Bush, 517 U. S., at 1003 (STEVENS, J., dissenting); id., at 1045 (SOUTER, J., dissenting); Shaw v. Hunt, 517 U. S. 899, 918 (1996) (STEVENS, J., dissenting). This suit exacerbates those concerns.

Legislators, for example, may ask just what the words "predominant racial motive" mean. The question has no obvious answer because racial motives (here efforts to include some additional African-American voters in a particular district) never explain a predominant portion of a district's entire boundary (most of which inevitably reflects county lines, other geographical features, and sometimes even a discriminatory history, see App. 120-121); yet those motives always predominate in respect to those voters (whether few or many) whom the legislature, with consciousness of race, places for that reason in one district rather than another. More importantly, here, unlike other cases that use somewhat similar words, the Court has not turned to other considerations, such as discriminatory intent, or vote dilution, or even a district's bizarre geographical shape, to help explain, or to limit the scope of, the words themselves. Cf. Shaw v. Hunt, supra; Regester, supra; Reynolds, supra; and Gomil

BREYER, J., dissenting

lion, supra. Thus, given today's suit, a legislator might reasonably wonder whether he can ever knowingly place racial minorities in a district because, for example, he considers them part of a "community" already there; because he thinks doing so will favor the Democrats (or the Republicans); because he wants to help an African-American incumbent; because he believes doing so will encourage participation in the political process by racial minorities in whom historical discrimination has induced apathy; because he believes that doing so will help those same voters secure representatives that better reflect their needs and desires; or simply because he wants to see more racial minorities elected to office in a Nation that has become increasingly diverse.

The Court has not said that the Constitution forbids the use of race in all these instances. See Adarand Constructors, Inc. v. Peña, 515 U. S. 200, 237 (1995); see also Shaw v. Reno, supra, at 646-647; Miller, supra, at 920; Bush, supra, at 1004 (STEVENS, J., dissenting); Wygant v. Jackson Bd. of Ed., 476 U. S., at 280; Richmond v. J. A. Croson Co., 488 U. S. 469, 493-494 (1989). If the use of race as a criterion is wrong in some, but not all, of these instances, the legislator will need to know when, and why. And the legislator will need a legal principle that tells him whether, or when, the answers to such questions vary depending upon whether the group is racial or reflects, say, economics, education, or national origin. Miller, supra, at 944-945 (GINSBURG, J., dissenting). It seems particularly difficult-without the use of some guiding or limiting principle, such as intent, vote dilution, or even bizarre district shape-to find principled legal answers to what, in the redistricting context, are traditionally political questions.

The decision also increases the risk of significant judicial entanglement in the inherently political redistricting process. See, e. g., Bush, supra, at 1035-1040 (STEVENS, J., dissenting); Miller, supra, at 934-935 (GINSBURG, J., dissenting); see also Growe v. Emison, 507 U. S. 25, 33-34 (1993);