dissenting.
Appellant seeks to stay the judgment of a three-judge Federal District Court, which held unconstitutional .Connecticut’s plan for apportioning its state legislature. 341 F. Supp. 139. The plan was adopted in September 1971, and was only in the preliminary stages of implementation when it was struck down as violative of the Equal Protection Clause on March 30, 1972. An appeal from that decision has been docketed in this Court. Gaffney v. Cummings, No. 71-1476.
We denied a motion for expedited consideration of that appeal on May 22, 1972. 406 U. S. 942. Appellant promptly moved the lower court for a stay of its *903March 30 decision, and when that stay was denied on May 26, 1972, appellant came here.
Earlier this Term, in another reapportionment case, Mr. Justice Powell; in an in-chambers opinion, articulated the considerations involved in our review of applications for a stay of lower court judgments:
“A lower court judgment, entered by a tribunal that was closer to the facts ... , is entitled to a presumption of validity. Any party seeking a stay of that judgment béars the burden of showing that the decision below was erroneous and that the implementation of the judgment pending appeal will lead to irreparable harm.” Graves v. Barnes, 404 U. S. 1201, 1203.
“Irreparable harm,” of course, inheres in any challenge to legislative apportionment. If the court below erred, the fall election will be held under an improper order, one which will doubtless affect the composition of the next state legislature. But this type of “irreparable injury” affects both sides equally, for if the court below was correct, staying its order will cause irreparable harm of precisely the same dimension.
There is “irreparable injury” in a different' sense if the court’s order striking down a state apportionment is handed down so near the upcoming election that it is administratively impractical to implement an orderly election. Here, there is no serious claim that'irreparable injury, in this sense, would result if a stay is not granted. The court below found as fact that there is ample time before the fall election' to implement the plan submitted by the Special Master on May 26, ‘1972, or any proposed substitute which the State or appellant might submit within a reasonable time.1 In*904deed, appellant concedes that the question of which plan can be most easily implemented is a “non-issue.”2
Thus, the issue determinative of the stay application is the probable correctness of the decision below, and, in my view, appellant has not met his burden “of showing that the. decision below was erroneous.”
In Reynolds v. Sims, 377 U. S. 533, 577, we said “the Equal Protection Clause requires that a State make an honest and good faith effort to construct districts, in both houses of its legislature, as nearly of equal population as is practicable.”. Moreover, a State may not be heard to argue that a population variance is justified because it is de minimis. “[T]he‘as nearly as practicable’ standard requires that the State make a good-faith effort to achieve precise mathematical equality. . . . Unless population variances among . . . districts are shown to have resulted despite such effort, the State must justify each variance, no matter how small.” Kirkpatrick v. Preisler, 394 U. S. 526, 530-531.3
A comparison of the population variances in this case with those disapproved in Kirkpatrick, supra, is striking. In Kirkpatrick, the average variation from the ideal district was only 1.6%. Here, assembly districts in the state plan exhibited an average variation of 1.9%. In Kirkpatrick, the ratio of the largest to the smallest district was only 1.06 to 1. Here, the- ratio of the largest to the smallest assembly district is 1.082 *905to 1. In Kirkpatrick, 70% of the districts were within plus or minus 1.88% of the ideal population figure. Here, only 51.65% of the assembly districts are within 2.0% of the ideal. In Kirkpatrick, the total variance4 was 5.97%. Here, the total variance of the assembly redistricting is 7.83%.
It is true, of course, that “the extent to which equality may practicably be achiéved may differ from State to State . . . ,” Kirkpatrick, supra, at 530. Thus, a State may be able to justify certain variations. Here, however, only two justifications are offered, and neither appears to have particular merit.
It is primarily argued that the variations are justified by a legitimate state interest in aphieving “a partisan balancing of strength in each house.” The District Court explained the concept as follows:.
“The partisan balancing of strength in each house, termed by intervening defendant [appellant in this Court] a ‘fair political balance’ and by plaintiffs [appellees herein] ‘political gerrymandering’ was obtained by so adjusting the census areas utilized as building blocks into the structuring of Senate and House districts that, on the basis of the vote for all the Senate candidates of each party in the elections of 1966, 1968 and 1970, whichever party carried the state should carry a majority of Senate seats proportional to the statewide party majority, and likewise in the House, based on the party vote for all the House candidates of each party in the same three elections.
“In one or more House and one or more Senate districts some accommodation was also made in *906the interest of retaining in office a particular incumbent/' 341 F. Supp. 139, 147.
This Court has never decided whether political gerrymandering or “fair political balance” is per se unconstitutional, irrespective of population variances. See, e. g., Wells v. Rockefeller, 394 U. S. 542, 544. But we have said, in no uncertain terms, that gerrymandering is not a justification where population variances do result. In Kirkpatrick, for example, we even rejected the State’s attempt to justify the population variances there present on the ground that the variations were necessary to avoid gerrymandering.
“[A]n argument that deviations from equality are justified in order to inhibit legislators from engaging in partisan gerrymandering is no more than a variant of the argument, already rejected, that considerations of practical politics can justify population disparities.” 394 U. S., at 534.
Thus, whether or not Connecticut may gerrymander its legislature if population equality is preserved, it may not do so when population disparities result.
An additional consideration urged to jústify the discrepancies is the State's interest in preserving town lines. But any weight this, factor would ordinarily have is rendered insignificant by the fact that the State’s own plan cuts across 47 towns to create assembly districts,, and 23 towns to create senate districts. See Whitcomb v. Chavis, 403 U.S. 124, 162 n. 42.
Appellant has one final argument. Attempting to litigate the merits of the Special Master’s plan, he argues that implementation of that plan would exceed the equity power of the federal court under our recent decision in Sixty-seventh Minnesota State Senate v. Beens, 406 U. S. 187. But the merits of the Special Master’s plan are not before this Court. Indeed, in *907denying the stay below, the District Court obligated itself to “set down for hearing with all reasonable dispatch the plan submitted by the Special Master and any other plans submitted/’ Whatever appellant’s objections to the Master’s plan might be, he should first air them in the District Court which stands ready to hear them.
Additionally,’ even were the Special Master’s plan at issue, appellant’s objections would not be well taken. This is not a case in which the size of a state house is “slashed” in half, as in Minnesota State Senate, supra. Here, the District Court merely reduced the size of Connecticut’s house from 151 members to 144, in order that the number of house districts be an even multiple of the 36 senate districts.5 A house of such size is expressly contemplated by the Connecticut Constitution.6 The District Court’s action is simply a “minor variation,” allowing senate and house districts to be drawn with congruent boundaries, that is well within the remedial powers of an equity court.7
I dissent from the Court’s order granting this stay.
The legislature has recently acted to remove what ever procedural roadblocks there might be to implementation of the Master’s plan *904or any other which the court below might adopt'. Public Act -220, May 16, 1972. The District Court indicated that the legislature will shortly submit a plan of its own for the court’s consideration. 341 F. Supp. 139, 150.
Appellant’s Reply Memorandum 4. See also Appellant’s Motion for Stay of Judgment 8.
It is irrelevant to this comparison that Kirkpatrick, 394 U. S. 526, involved congressional rather than state legislative redistricting. In either case, the burden is on the State to demonstrate a valid justification for any population variance, no matter how small.
The “total variance” in an apportionment plan is derived by adding together the percentage variation from the ideal of the two districts which are respectively the most overpopulated and underpopulated.
Minor variations for this purpose were approved in the Minnesota State Senate case. 406 U. S. 187, ¿nd cases cited id., at 198 n. 10.
Article III, § 4, of the Connecticut Constitution provides that “The house of representatives shall consist of not less than one hundred twenty-five and not more than two hundred twenty-five members . . .
Appellant also objects to the extent to which the:Master’s plan, dishonors town boundaries. It is undisputed, however, that town boundaries cannot be preserved intact in all cases under any constitutional plan. The Master’s plan, drawn with the preservation of as many town lines as possible as an express consideration (though a subordinate one to the goal of population equality), cuts across only 60 towns- in creating assembly districts, and 30 towns in creating senate districts. These figures compare favorably with those in the State’s plan, supra, at 906.