delivered the opinion of the Court.
This reapportionment case was brought in the District Court of Hawaii by residents and qualified voters of the City and County of Honolulu, appellees in each of the three appeals consolidated here. They alleged that Hawaii’s legislative apportionment, was unconstitutional under our decisions in Reynolds v. Sims, 377 U. S. 533, and companion cases.1 William S. Richardson, Lieutenant Governor of Hawaii, also an appellee in all three appeals, was named defendant in his capacity as the state officer responsible for supervising state elections. John A. Burns, Governor of Hawaii, appellant in No. 318, intervened as a party plaintiff. Members of the State *76House of Representatives, appellants in No. 323, and members of the State Senate, appellants in No. 409, intervened as parties defendant.
Under the Hawaii Constitution, adopted in 1950 and put into effect upon admission to statehood in 1959, the State is divided into four major counties, referred to in the State Constitution as “basic areas.” Each county is made up of a group of islands, separated from each of the other counties by wide and deep ocean waters. The principal island of the City and County of Honolulu, the most populous county, is the island of Oahu. It is the State’s industrial center, principal tourist attraction, and site of most of the many federal military establishments located in the State. In 1960, 79% of the State’s population lived there. The three other counties, primarily rural and agricultural, are Hawaii County, Maui County, and Kauai County.2
The apportionment article of the State Constitution was framed to assure that the three small counties would choose a controlling majority of the State Senate and that the population center, Oahu, would control the State House of Representatives. Thus, Art. Ill, § 2, of the State Constitution apportions a 25-member senate among six fixed senatorial districts, assigning a specified number of seats to each. Fifteen senate seats, a controlling majority, are allocated among Hawaii, Kauai and *77Maui Counties and 10 seats are assigned to Oahu. Alteration of this apportionment is made very difficult by a provision “that no constitutional amendment altering . . . the representation from any senatorial district in the senate shall become effective unless it shall also be approved by a majority of the votes tallied upon the question in each of a majority of the counties.” 3 Hawaii Const., Art. XV, § 2, ¶ 6.
For the State House of Representatives, on the other hand, the State Constitution establishes 18 representative districts, 10 of which are on Oahu, and requires the Governor to apportion the 51-member body among these districts on the basis of the number of voters registered in each. The first apportionment occurred in 1959, just prior to statehood, and was based on registration figures for the 1958 territorial election. It produced 13 multi-member representative districts and five single-member districts, and allocated 36 representatives, a controlling majority, to Oahu.4 The Governor is required to reap*78portion the State decennially, a duty which may be enforced by mandamus from the State Supreme Court.
This apportionment scheme was first attacked in the Supreme Court of Hawaii, within a month after we decided Reynolds v. Sims. That court refused to pass on the validity of the apportionment at that time. It noted the imminence of the 1964 election and stated its belief that, consistent with the Hawaii Constitution, judicial proceedings should await legislative proposals for a constitutional amendment or a constitutional convention. Guntert v. Richardson, 47 Haw. 662, 394 P. 2d 444. Compare Reynolds v. Sims, 377 U. S., at 585. A special legislative session was then called by the Governor to consider reapportionment. It failed to act.
This suit was brought on August 13, 1964. A three-judge court was convened, as required by 28 U. S. C. §§2281, 2284 (1964 ed.). Interim relief was denied in view of the pendency of the 1964 elections and hearings were set for January 1965. The court published its first decision and order on February 17, 1965. 238 F. Supp. 468. That order declared all provisions of the apportionment plan contained in the Hawaii Constitution valid under the Equal Protection Clause except the mentioned provisions relating to the apportionment of the State Senate. These were affirmatively declared to be invalid and unconstitutional.
In the February 17 order the District Court decided not to fashion its own reapportionment plan for the senate. Nor did it instruct the legislature to reapportion the senate or to propose constitutional amendments for that purpose.5 Instead, it directed the legislature to sub*79mit to the electorate at an immediate special election the question, “Shall there be a convention to propose a revision of or amendments to the Constitution?” The legislature was also directed to establish the convention procedures according to a timetable the court set.6 The court retained jurisdiction for all purposes, including that of itself reapportioning the senate in the event of *80a negative vote on the question, failure of the convention to adopt a suitable amendment, or rejection by the electorate of the amendment adopted by the convention.
The court chose the convention route over the legislative route for two reasons. Under the Hawaii Constitution all elections necessary to adoption of amendments proposed by a constitutional convention may be held on a special basis. Legislative proposals, on the other hand, may be submitted only at a general election. In starting the machinery necessary for a convention, the court hoped that a valid permanent plan could be presented to the electorate and adopted before the next general election, to be held in 1966. The second reason was that the court doubted that the legislature would be able to agree on an amendment proposal for reapportioning the senate, in view of the failure of the previously called legislative special session to act.
The special elections necessary under the court’s order, however, entailed substantial expense. On motion of the intervening legislators, which showed substantial progress towards a legislative proposal for amendment, the court on March 9,1965, modified its order. As suggested by the parties, it suspended the February 17 order and instead required the legislature to enact three separate statutes before turning to regular legislative business. One statute was to propose an interim senate apportionment plan, using registered voters as a basis, to be submitted to the court. If approved, it would be adopted by the court as its plan for use in the 1966 general election. The second statute was to propose a constitutional amendment embodying pertinent provisions of the interim plan, to be submitted to the people for approval at that election. The third statute was to submit the question of calling a constitutional convention to the electorate at the 1966 general election.
*81Three statutes were enacted. H. B. 987, the only one of these measures before us,7 proposed an interim plan of apportionment for the senate. 1 Hawaii Sess. Laws 1965, Act 281. The plan followed the pattern for house apportionment. It established eight senatorial districts, five on Oahu. As required by the court’s order, the 25 senators were to be apportioned on the basis of registered voters.8 Using figures derived from registration for the 1964 general elections, Oahu was allocated 19 out of the 25 senators, a controlling majority.
Under the total apportionment scheme which resulted from this enactment, Oahu would not have any single-member districts in either the house or the senate. The distribution of registered voters in Oahu is such that Oahu’s 10 representative districts have two to six representatives each, and its five senatorial districts each would have either three or four senators. Hawaii County would be a single senatorial district represented by three senators and have five representative districts, four choosing a single representative and the fifth electing three. Maui County would be a single senatorial district electing two senators and have two representative districts, one electing four, and the other a single representative. *82Kauai County would be a single senatorial and a single representative district electing one senator and three representatives. Thus, Oahu with 79% of total population would elect 76% of the senate, 19 of 25 senators, and 71% of the house, 36 of 51 representatives.
The new senate apportionment scheme was submitted to the court immediately upon passage. By opinion and order of April 28, 1965, the District Court disapproved it, and reinstated the provision of its earlier order requiring immediate resort to the convention method.9 240 F. Supp. 724. It expressly approved the use of the registered voters measure of population. Its disapproval was based on the legislative decison not to create single-member senatorial districts for Oahu but merely to increase the number of multi-member senatorial districts on that island from two to five. It was not contended that the apportionment failed to meet the standard of Reynolds v. Sims if the use of multi-member districts and the use of registered voters as the apportionment base did not offend the Equal Protection Clause.10
In May 1965, the Governor filed a notice of appeal to this Court from certain provisions of the two orders and thereafter the participating senators and representatives also filed notices of appeal from parts of the orders.11 *83We noted probable jurisdiction and consolidated the appeals for argument. 382 U. S. 807. We set aside and vacate both orders and remand for further proceedings consistent with this opinion.
I.
All parties concede the invalidity of the provisions of Art. Ill, § 2, apportioning the senate on the basis of geography rather than population, and of the provision of Art. XV, § 2, ¶ 6, requiring a majority vote of the electorate in each of a majority of the counties to amend senatorial apportionment established by the constitution. The District Court concluded that, as a matter of state law, the house and senate apportionment plans were severable. Compare Lucas v. Colorado General Assembly, 377 U. S. 713, 735. Even so, Maryland Committee v. Tawes, 377 U. S. 656, holds that a court in reviewing an apportionment plan must consider the scheme as a whole. Implicit in this principle is the further proposition that the body creating an apportionment plan in compliance with a judicial order should ordinarily be left free to devise proposals for apportionment on an overall basis. The Governor argues that the District Court committed “fundamental error” in preventing the Hawaii Legislature from engaging in such deliberations, and that for that reason alone the legislative product was inevitably tinged with constitutional error.
We agree that, once the District Court decided to permit legislative action, it could and should have made clear to the Hawaii Legislature that it could propose modification of the house as well as the senate plan, both as to the interim apportionment to be adopted under *84court order and as to proposals for permanent reapportionment through constitutional amendment. That approach would have enabled the legislature to channel its efforts to permanent rather than temporary change. Indeed, the failure to invite such thoroughgoing consideration was particularly unfortunate in connection with the court’s requirement in its order of March 9 that the Hawaii Legislature prepare constitutional amendments for a permanent apportionment plan. By directing that the permanent plan incorporate the interim reapportionment plan and by restricting the choices available to the legislature in adopting an interim plan, the court put significant restraints on the legislature’s deliberations about permanent apportionment. It seemed not only to limit the legislature to consideration of senate apportionment but also to require that a registered voters basis be used for that apportionment. These constraints, together with the District Court’s action in explicitly sustaining the constitutionality of the house apportionment in its order of February 17, may have limited the opportunities of the legislature to construct the total scheme of apportionment best suited to the State’s needs.12 Our decision in Reynolds v. Sims emphasized *85that “legislative reapportionment is primarily a matter for legislative consideration and determination, and that judicial relief becomes appropriate only when a legislature fails to reapportion according to federal constitutional requisites in a timely fashion after having had an adequate opportunity to do so.” 377 U. S., at 586. Until this point is reached, a State’s freedom of choice to devise substitutes for an apportionment plan found unconstitutional, either as a whole or in part, should not be restricted beyond the clear commands of the Equal Protection Clause.
We are dealing here, however, only with the interim plan. The State remains free to adopt other plans for apportionment, and the present interim plan will remain in effect for no longer time than is necessary to adopt a permanent plan. The 1966 general elections are imminent, and election machinery must be put into operation before further proceedings could be completed. In this context, the question of embarrassment of state legislative deliberations may be put aside. For present purposes, H. B. 987 may be treated together with the existing house apportionment as a new, overall proposal for interim apportionment. The only question for us is whether, viewing the resulting plan in its en*86tirety and without regard to its history, it falls short of federal constitutional standards. We conclude for reasons to be stated that H. B. No. 987 and the existing house apportionment together constitute an interim legislative apportionment which has not been shown to fall short of federal standards. We direct the District Court to enter an order appropriate to adopt the plan as the court’s own for legislative apportionment applicable to the 1966 election, and thereafter until a constitutional permanent plan is adopted, constitutional deficiencies in the interim plan are shown, or another interim plan for reapportionment of the Hawaii Legislature suggested by the legislature is approved by the court.
II.
The April 28 opinion began analysis in terms of the interim senate apportionment plan’s effect upon representation in the State’s scheme of representation as a whole. The District Court was not concerned with population disparities, however, but with what it considered to be a difference in representational effectiveness between multi-member and single-member legislative districts.13 In an informal memorandum circulated among *87the parties in early April, the District Court had advised the legislature of its doubts concerning the validity of a multi-member senatorial districting plan, saying:
“We believe that the Senate should be redistricted into single senatorial districts, although we may approve two-member districts if and only if the legislature can affirmatively show substantial reasons therefor. There may very well be valid reasons for one or two 2-member districts in the neighboring islands but we perceive no justification whatsoever for other than single member districts on the Island of Oahu, particularly the heavily populated areas thereof.”
The opinion of April 28 clearly reveals that the court was still convinced that only single-member senatorial dis-tricting on Oahu would be appropriate. It felt, for example, that the legislature had “built monoliths” into the districting scheme by making the boundaries of the third senatorial district and the eighth representative district one and the same, thus enabling the same constituency to elect four representatives and three senators, and by fashioning the sixth senatorial district almost entirely from the fifteenth representative district, from which six representatives and four senators would be elected. It also felt that in setting up the senatorial districts on Oahu the legislature had not taken into account “community of interests, community of problems, socio-economic status, political and racial factors”; and, finally, that “the legislature’s adamant insistence on three and four-member senatorial districting was the conscious or unconscious— though not unnatural — reluctance of the affected senators to carve out single-member districts which thereafter would in all probability result in a political duel-to-the-death with a fellow and neighbor senator.” 240 F. Supp., at 730-731.
*88But the Equal Protection Clause does not require that at least one house of a bicameral state legislature consist of single-member legislative districts. See Fortson v. Dorsey, 379 U. S. 433. Where the requirements of Reynolds v. Sims are met, apportionment schemes including multi-member districts will constitute an invidious discrimination only if it can be shown that “designedly or otherwise, a multi-member constituency apportionment scheme, under the circumstances of a particular case, would operate to minimize or cancel out the voting strength of racial or political elements of the voting population.” Id., at 439.
It may be that this invidious effect can more easily be shown if, in contrast to the facts in Fortson, districts are large in relation to the total number of legislators, if districts are not appropriately subdistricted to assure distribution of legislators that are resident over the entire district, or if such districts characterize both houses of a bicameral legislature rather than one. But the demonstration that a particular multi-member scheme effects an invidious result must appear from evidence in the record. Cf. McGowan v. Maryland, 366 U. S. 420. That demonstration was not made here.14 In relying on conjecture as to the effects of multi-member districting rather than demonstrated fact, the court acted in a manner more appropriate to the body responsible for drawing up the districting plan. Speculations do not supply evidence that the multi-member districting was designed to have or had the invidious effect necessary to a judgment of *89the unconstitutionality of the districting. Indeed, while it would have been better had the court not insisted that the legislature “justify” its proposal, except insofar as it thus reserved to itself the ultimate decision of constitutionality vel non, the legislature did assign reasons for its choice.15 Once the District Court had decided, properly, not to impose its own senate apportionment but to allow the legislature to frame one, such judgments were exclusively for the legislature to make. They were subject to constitutional challenge only upon a demonstration that the interim apportionment, although made on a proper population basis, was designed to or would operate to minimize or cancel out the voting strength of racial or political elements of the voting population.16
*90III.
The dispute over use of distribution according to registered voters as a basis for Hawaiian apportionment arises because of the sizable differences in results produced by that distribution in contrast to that produced by the distribution according to the State’s total population, as measured by the federal census figures. In 1960 Oahu’s share of Hawaii’s total population was 79%. Its share of persons actually registered was 73%. On the basis of total population, Oahu would be assigned 40 members of the 51-member house of representatives; on the basis of registered voters it would be entitled to 37 representatives.17 Probably because of uneven distribution of military residents — largely unregistered — -the differences among various districts on Oahu are even more striking. For example, on a total population basis, Oahu’s ninth and tenth representative districts would be entitled to 11 representatives, and the fifteenth and sixteenth representative districts would be entitled to eight. On a reg*91istered voter basis, however, the ninth and tenth districts claim, only six representatives and the fifteenth and sixteenth districts are entitled to 10.18
The holding in Reynolds v. Sims, as we characterized it in the other cases decided on the same day, is that “both houses of a bicameral state legislature must be apportioned substantially on a population basis.” 19 We start with the proposition that the Equal Protection Clause does not require the States to use total population figures derived from the federal census as the standard by which this substantial population equivalency is to be measured. Although total population figures were in fact the basis of comparison in that case and most of the others decided that day, our discussion carefully left open the question what population was being referred to. At several points, we discussed substantial equivalence in terms of voter population or citizen population, making no distinction between the acceptability of such a test and a test based on total population.20 Indeed, in WMCA, Inc. v. Lomenzo, 377 U. S. 633, decided the same day, we treated an apportionment based upon United States citizen population as presenting problems *92no different from apportionments using a total population measure. Neither in Reynolds v. Sims nor in any other decision has this Court suggested that the States are required to include aliens, transients, short-term or temporary residents, or persons denied the vote for conviction of crime, in the apportionment base by which their legislators are distributed and against which compliance with the Equal Protection Clause is to be measured.21 The decision to include or exclude any such group involves choices about the nature of representation with which we have been shown no constitutionally founded reason to interfere. Unless a choice is one the Constitution forbids, cf., e. g., Carrington v. Rash, 380 U. S. 89, the resulting apportionment base offends no constitutional bar, and compliance with the rule established in Reynolds v. Sims is to be measured thereby.
Use of a registered voter or actual voter basis presents an additional problem. Such a basis depends not only upon criteria such as govern state citizenship, but also upon the extent of political activity of those eligible to register and vote. Each is thus susceptible to improper influences by which those in political power might be able to perpetuate underrepresentation of groups constitutionally entitled to participate in the electoral process, *93or perpetuate a “ghost of prior malapportionment.” 22 Moreover, “fluctuations in the number of registered voters in a given election may be sudden and substantial, caused by such fortuitous factors as a peculiarly controversial election issue, a particularly popular candidate, or even weather conditions.” Ellis v. Mayor & City Council of Baltimore, 352 F. 2d 123, 130 (C. A. 4th Cir. 1965).23 Such effects must be particularly a matter of concern where, as in the case of Hawaii apportionment, registration figures derived from a single election are made controlling for as long as 10 years. In view of these considerations, we hold that the present apportionment satisfies the Equal Protection Clause only because on this record it was found to have produced a distribution of legislators not substantially different from that which would have resulted from the use of a permissible population basis.
As the District Court noted, the 1950 constitutional convention discussed three possible measures, total population, state citizen population, and number of registered voters, in considering how the State House of Representatives should be apportioned. Apportionment under the Organic Act had been on the basis of citizen population; this had proved difficult to administer because statistics were not readily available. Total population was disfavored because the census tracts, by which it is determined and reported, did not necessarily comport with traditional local boundaries. Registered voters was chosen as a reasonable approximation of both citizen and total population — readily determinable, conveniently *94broken down by election district, and a measure which, as against total population, somewhat favored the other islands over Oahu. It-is fair to say that the convention report reflected that citizen population as much as total population was the basis against which a registered voters standard was compared.
Hawaii’s special population problems might well have led it to conclude that state citizen population rather than total population should be the basis for comparison. The District Court referred to the continuing presence in Hawaii of large numbers of the military: “Hawaii has become the United States’ military bastion for the entire Pacific and the military population in the State fluctuates violently as the Asiatic spots of trouble arise and disappear. If total population were to be the only acceptable criterion upon which legislative representation could be based, in Hawaii, grossly absurd and disastrous results would flow . . . .” 238 F. Supp., at 474.24 Similarly, the court referred to the distortion in census figures attributable to “the large number of tourists who continually flow in and out of the State and who ... for census purposes are initially at least, counted as part of Hawaii’s census population . . . .” Id., at 475. (Footnote omitted.) Both the tourists and the military tend to be highly concentrated on Oahu and, indeed, are largely confined to particular regions of that island. Total population figures may thus constitute a substantially distorted reflection of the distribution of state citizenry. If so, a finding that registered voters distribution does *95not approximate total population distribution is insufficient to establish constitutional deficiency. It is enough if it appears that the distribution of registered voters approximates distribution of state citizens or another permissible population base.
Because state citizen population figures are hard to obtain or extrapolate, a comparison of the results which would be obtained by use of such figures with the results obtained by using registered voter figures is difficult. But the District Court found that military population of Oahu, and its distribution over that island, was sufficient to explain the already noted differences between total population and registered voters apportionments, both as among Hawaii’s four counties and as among Oahu’s representative districts. The District Court noted “that there is nothing in the State Constitution or the Hawaii statutes which per se excludes members of the armed forces from establishing their residence in Hawaii and thereafter becoming eligible to vote. This court finds no scheme in Hawaii’s Constitution or in the statutes implementing the exercise of franchise which is aimed at disenfranchising the military or any other group of citizens.” 238 F. Supp., at 475. No issue was raised in the proceedings before it that military men had been excluded improperly from the apportionment base.25 *96Moreover, the District Court stressed that Hawaii’s Constitution and laws actively encourage voter registration. A high proportion of the possible voting population is registered,26 and “strong drives to bring out the vote have resulted in a vote of from 88 to 93.6% of all registered voters during the elections of 1958, 1959, 1960 and 1962.” Id., at 476 (footnote omitted). In these circumstances, we find no demonstrated error in the District Court’s conclusion that the apportionment achieved by use of a registered voters basis substantially approximated that which would have appeared had state citizen population been the guide.
We are not to be understood as deciding that the validity of the registered voters basis as a measure has been established for all time or circumstances, in Hawaii or elsewhere. The District Court was careful to disclaim any holding that it was a “perfect basis.” We agree. It may well be that reapportionment more frequently than every 10 years, perhaps every four or eight years, would better avoid the hazards of its use. Use of presidential election year figures might both assure a high level of participation and reduce the likelihood that varying degrees of local interest in the outcome of the election would produce different patterns of political activity over *97the State. Other measures, such as a system of permanent personal registration, might also contribute to the stability and accuracy of the registered voters figure as an apportionment basis. Future litigation may reveal infirmities, temporary or permanent, not established by the present record.27 We hold that, with a view to its interim use, Hawaii’s registered voter basis does not on this record fall short of constitutional standards.
IV.
Our conclusion that the interim apportionment should apply to the 1966 election requires that the provisions of the order of February 17 mandating an immediate special election on the question of calling a constitutional convention should remain inoperative. The imminence of the 1966 elections precludes any further action pending that event. But the question remains what role the District Court has in bringing about a permanent reapportionment as promptly as reasonably may be after that election. We believe it should retain jurisdiction of the case to take such further proceedings as may be appropriate in the event a permanent reapportionment is not made effective. We note that the electorate will vote at the 1966 election on the question whether a constitutional convention should be convened. We see no reason, however, why the newly elected legislature should either be compelled to propose amendments or be precluded from proposing them. The legislature will doubt*98less find reason enough to act in the fact that the District Court will retain jurisdiction over the cause to take any action that may be appropriate pending the adoption of a permanent reapportionment which complies with constitutional standards. Such action may include further inquiry into the constitutionality of the present plan in its operation, consideration of substitute interim plans for apportioning the house and senate that might be submitted by the legislature in the event of failure of proposals for constitutional amendment, or judicial apportionment if the present plan is shown to be constitutionally deficient and no acceptable substitute is forthcoming.
The District Court is accordingly directed on remand to enter an appropriate order (1) adopting H. B. No. 987 and the existing house apportionment as an interim legislative apportionment for Hawaii and (2) retaining jurisdiction of the cause for all purposes.
Our judgment shall issue forthwith.
Vacated and remanded.
Mr. Justice Fortas took no part in the consideration or decision of this case.WMCA, Inc. v. Lomenzo, 377 U. S. 633; Maryland Committee v. Tawes, 377 U. S. 656; Davis v. Mann, 377 U. S. 678; Roman v. Sincock, 377 U. S. 695; and Lucas v. Colorado General Assembly, 377 U. S. 713.
Kalawao, a Hansen’s disease treatment area, is considered a fifth county for some purposes. However, its residents are considered part of Maui County for political purposes, and vote in that county for state legislators. We therefore treat only the four major counties, or basic areas, in this opinion. The State’s 1960 population of 632,772 was divided among these four counties as follows: City and County of Honolulu, 500,409; Hawaii County, 61,332; Maui County, 42,855; and Kauai County, 28,176. The population of the small, outlying islands other than Oahu which comprise the City and County of Honolulu is negligible. We therefore refer to that county hereafter as Oahu.
The District Court found that “this proviso was specifically inserted in order to freeze representation in the senate, and it gave to the rural counties what amounted to the right of veto over any attempt to change the representative makeup of the senate.” 238 F. Supp. 468, 472.
Hawaii uses the method of equal proportions to distribute legislators, first among the four counties and then among the districts within each county. This is the same method as used in apportioning the members of the House of Representatives of the United States Congress. Complex mathematically, it determines a priority order in which legislators are to be assigned among various competing districts. The system is discussed in Schmeckebier, The Method of Equal Proportions, 17 Law & Contemp. Prob. 302 (1952). Use of this method will not necessarily result in a constitutional apportionment. It is the distribution of legislators rather than the method of distributing legislators that must satisfy the demands of the Equal Protection Clause. No claim is made, however, that the effect of applying the method in Hawaii in this case was to deny any person equal protection of the laws by creating representative districts substantially unequal in size.
The court doubted whether the legislature itself had authority under state law to adopt an interim apportionment plan, in view of the decision in Guntert v. Richardson, supra. The Hawaii Constitution authorizes the legislature to propose constitutional amendments to the electorate either upon passage by a two-thirds vote of *79both houses of the legislature or upon passage by a majority vote of both houses in each of two successive legislative sessions. The Hawaii Constitution also authorizes the legislature to submit to the people the question of calling a constitutional convention, either at a general election or at a special election called for that purpose. Hawaii Const., Art. XV, § 3.
Paragraph 4 of the court’s order provided:
“4. This court will not interfere with the convening or conducting of the business of the Third State Legislature in regular session in 1965, save and except that the parties herein are hereby enjoined from taking final action upon any legislation, except such actions as are necessary to organize the respective houses at such session and appropriate funds for the session, until legislation, pursuant to the provision of Article XV of said Constitution providing for the submission to the people of Hawaii, by special election to be held not later than August 1, 1965, the question: 'Shall there be a convention to propose a revision of or amendments to the Constitution?’, and for any and all acts required by law to implement such legislation, has been enacted into law. Such legislation shall also provide that if the vote be in the constitutional affirmative, then a special election shall be held not later than September 15, 1965 to elect delegates to the convention in the manner provided in the Constitution. Such legislation may include legislative action under Article XV, Section 2, 4th paragraph, of the Constitution. Such legislation shall further provide that the convention convene not later than October 15, 1965 and that it conclude its deliberation in time to submit its proposed constitutional amendments to the electorate of Hawaii at a special election to be held not later than January 30, 1966, including (but not limiting the convention thereto) provisions therein for reapportioning the Senate of Hawaii on a constitutionally valid basis. Such legislation shall also appropriate and make available funds for the expenses of such elections and convention.” 238 F. Supp., at 479.
H. B. 986, 1 Hawaii Sess. Laws 1965, Act 280, provides for submission to the electorate in the 1966 general election of the question whether a constitutional convention should be called. H. B. 773, 1 Hawaii Sess. Laws 1965, p. 483, proposing a constitutional amendment in the same form as the interim plan, was passed by only a majority vote in the senate and hence must be acted on again before it can be submitted to the people for adoption or rejection. See n. 5, supra. In view of the constraints placed on the legislature in adopting this proposal, we think the District Court on remand should make no attempt to require any further action on this measure. See Part I, infra.
The method of equal proportions was to be used for apportioning the senate as well as the house. See n. 4, supra.
On May 21, 1965, Me. Justice Douglas stayed this action pending our determination of these appeals.
We are not to be understood as agreeing with the District Court, insofar as it may have rested its decision on the view that use of the method of equal proportions itself saved the plan from constitutional challenge based on Reynolds v. Sims. 240 F. Supp., at 727. See n. 4, supra.
These notices were timely filed. The February 17 opinion was not formally entered until April 9, 1965. The second decision was dated and entered April 28, 1965. Notices of appeal were filed May 3 and 7, 1965. Whether judged by the date of entry, United States v. Hark, 320 U. S. 531; Fed. Rule Civ. Proc. 58, or by the fact that the order incorporated in the decision of February 17 was *83not finally made effective until the decision of April 28, United States v. Crescent Amusement Co., 323 U. S. 173, 177, the appeals from the decision announced February 17 were timely. 28 U. S. C. § 2101(b) (1964 ed.).
We have not overlooked the fact that the limitations were suggested by the legislators themselves. Nevertheless, consistently with Maryland Committee v. Tawes, the District Court should have indicated to the legislators that they possessed the same broad scope of inquiry as the court had said in its opinion of February 17 was open to a constitutional convention. It had suggested there that there might be appropriately considered by the convention:
“1. Whether [Hawaii] will continue to use registered voters as the apportionment basis, or change it to State citizen population eligible to vote (i. a., voter population), or citizen population, or total population.
“2. Whether it is better to have one or both houses of the legislature composed of single member representative districts, or to have *85and justify one or both houses composed, in whole or in part, of multi-member or floterial districts.
“3. Whether decennial reapportionment of either or both houses should be made on or before June 1st of the year preceding the Federal census — as is now the case — or on a date soon after the taking of such census.
“4. Whether the representative district lines should remain substantially as they now are or whether ultimately (i. e., after 1970) there should be redistricting in such a manner that the census tracts and representative districts can be coordinated for the statistical purposes necessary to implement the changes (if any) made in the basis of reapportionment.” 238 F. Supp., at 478.
“In reapportioning and redistricting the senate, both houses overlooked the fact that, to be valid, the makeup of the senate must positively complement the makeup of the house, to provide the vital equality of voter representation. Both houses of the legislature seemingly forgot that the schemes of districting each house, when conjoined, must offer compensating advantages to the voters — not only to those voters within each representative district, be it senate or house, but to all voters throughout the State. While there perforce must be some overlap of representation with the several senate and house districts, that overlap must not be such as to concentrate and intensify the voting power of a single senatorial-representative district to the point that the voters therein have a built-in disproportionate representational advantage over any other voters of the State.” 240 F. Supp., at 729.
Appellant Burns concedes in his brief that “[i]n the case of the Hawaii House multimember districts, extensive proofs were not put in as to the details of the submergence of minorities.” There may, for example, be merit in the argument that by encouraging block voting multi-member districts diminish the opportunity of a minority party to win seats. But such effects must be demonstrated by evidence.
As stated in the court’s opinion, the legislature’s proffered justifications were:
“(1) single-member districts would tend to cause the senators therefrom to be concerned with localized issues and ignore the broader issues facing the State, and therefore it might fragment the approach to state-wide problems and programs to the detriment of the State; (2) historically the members of the house had represented smaller constituencies than members of the senate, and tradition and experience had proved the balance desirable; (3) multi-member districts would increase the significance of an individual’s vote by focusing his attention on the broad spectrum of major community problems as opposed to those of more limited and local concern; (4) to set up single-member districts would compound the more technical and more intricate problem of drawing the boundaries; (5) population shifts would more drastically affect the boundaries of many' smaller single-member districts — to a greater degree than would be found in larger multi-member districts, citing Oahu’s population boom and subdivision development.” 240 F. Supp., at 727.
We reject the suggestion that the districts are arbitrarily or invidiously defined. The fact that district boundaries may have been drawn in a way that minimizes the number of contests between present incumbents does not in and of itself establish invidiousness. And we find no support for this suggestion in the present wide variances *90in size among the Oahu representative districts. This distribution is governed by the population shifts which have occurred since the district boundaries were first defined. In the initial apportionment, the six representative districts comprising the fifth senatorial district each contained two or three representatives — two in the geographically large, relatively rural districts and three in urban districts. The four representative districts comprising the fourth senatorial district contained three to six representatives; these districts comprised the heart, of residential Honolulu, and were understandably compact. Whether one surmises that the drafters were leaving room for expansion in the less populous districts or drawing district lines as a function of size as well as population, no irrationality appears from the distribution. It is relevant to note that the Hawaii Legislature was dominated by multi-member districts in both houses before statehood. This feature thus did not originate with the senate plan here under consideration.
This figure is calculated using 1960 figures; in the apportionment of 1959 Oahu was assigned 36 representatives, on the basis of 1958 registration figures.
Thus, in 1960, the ninth and tenth districts contained 28% of Oahu’s population but only 17% of its registered voters; the fifteenth and sixteenth districts, with only 21% of island population contained 29% of island registered voters.
E. g., WMCA, Inc. v. Lomenzo, 377 U. S., at 653; Maryland Committee v. Tawes, 377 U. S., at 674.
Thus we spoke of “[t]he right of a citizen to equal representation and to have his vote weighted equally with those of all other citizens . . . .” Reynolds v. Sims, 377 U. S., at 576. We also said: “[I]t is a practical impossibility to arrange legislative districts so that each one has ah identical number of residents, or citizens, or voters.” Id., at 577. “[T]he overriding objective must be substantial equality of population among the various districts, so that the vote of any citizen is approximately equal in weight to that of any other citizen in the State.” Id., at 579.
In Davis v. Mann, 377 U. S., at 691, we rejected an argument that underrepresentation of three political subdivisions in Virginia was “constitutionally justifiable since it allegedly resulted in part from the fact that those areas contain large numbers of military and military-related personnel. Discrimination against a class of individuals, merely because of the nature of their employment, without more being shown, is constitutionally impermissible.” See also Carrington v. Rash, 380 U. S. 89. Where the exclusion is of those not meeting a State’s residence requirements, however, different principles apply. The difference between exclusion of all military and military-related personnel, and exclusion of those not meeting a State’s residence requirements is a difference between an arbitrary and a constitutionally permissible classification.
Buckley v. Hoff, 243 F. Supp. 873, 876 (D. C. Vt. 1965).
Ellis disapproved a registered voters basis for apportioning the governing council of Baltimore, Maryland. The Court of Appeals held that this basis was permissible only if it yielded results substantially approximating those obtained by use of a total population base.
For example, at one point during World War II, the military population of Oahu constituted about one-half the population of the Territory. If total population were used in such a situation, the permanent residents living in districts including military bases might have substantially greater voting power than the electors of districts not including such bases. Indeed, in view of this possibility, appellant Burns concedes that a “nontransient” figure as well as total population might be used for apportionment purposes.
Appellant Burns urges here that the apportionment base for the house, registered voter figures from the 1958 general election, is infected by such an exclusion. Hawaii was then a Territory, and registration was governed by 48 U. S. C. §619 (1958 ed.), which provided: “No person shall be allowed to vote who is in the Territory by reason of being in the Army or Navy or by reason of being attached to troops in the service of the United States.” Such a restriction, if imposed by a State, would violate the Equal Protection Clause. Carrington v. Rash, 380 U. S. 89. The statute no longer applies, but its effect persists in the house apportionment. The number of registered voters in the districts where Oahu’s major military bases are located has increased twice as much as reg*96istration in the other Oahu districts and more than three times as much as state population since 1958. Reapportionment of the house now on a registered voters basis would work a substantial realignment of the State’s representative districts. If it can be shown that this is so principally because military men now have a vote they were once denied, rather than because of simple population shifts, an immediate interim adjustment of house apportionment might be merited. Time does not permit the necessary hearings to be had before the 1966 elections, but requiring such hearings is certainly within the court’s authority under its continuing jurisdiction thereafter.
The District Court found the figure to be 87.1%. Even if an asserted error in statistics is corrected, the figure exceeds 80%.
Note 25, supra. An attempt was made to show that registration percentages among low-income residents of Oahu were substantially lower than among other resident groups. It is unclear to what extent these statistics reflect military pay scales. Thus, they may be an unfair representation of state citizen registration patterns. Moreover, no substantial effect in submerging the political voice of this group appears. Of course, this issue may be re-examined should further hearings be held in exercise of the court’s continuing jurisdiction.