Doulin v. White, 528 F. Supp. 1323 (E.D. Ark. 1982)
January 5, 1982
Frank WHITE, Governor of the State of Arkansas; Winston Bryant, Lieutenant Governor of the State of Arkansas; Steve Clark, Attorney General of the State of Arkansas; Paul Riviere, Secretary of State of the State of Arkansas; Julia Hughes Jones, State Auditor of the State of Arkansas; Jimmie Lou Fisher, State Treasurer of the State of Arkansas; Bill McCuen, State Land Commissioner of the State of Arkansas; Herby Branscum, Chairman of the State Democratic Central Committee of the State of Arkansas; Harlan "Bo" Holleman, Chairman of the State Republican Central Committee of the State of Arkansas; Joe Basore, Citizen of the State of Arkansas; E. J. Jacobs, Citizen of the State of Arkansas; Stark Ligon, Citizen of the State of Arkansas; all of the above being Commissioners of the State Board of Elections of the State of Arkansas, Defendants;
Veo Easley, County Judge of Grant County, Arkansas, Plaintiff-Intervenor.
United States District Court, E. D. Arkansas, W. D.
*1324 Robert Walker, Heiskell, Donelson, Bearman, Adams, Williams, & Kirsch, John L. Ryder, Laughlin, Halle, Clark & Gibson, Memphis, Tenn., Stan Miller, Miller, Jones & Goldman, Hot Springs, Ark., for plaintiffs.
R. B. Friedlander and Debbie Nye, Asst. Attys. Gen., Little Rock, Ark., for defendants.
Phillip H. Shirron, Sheridan, Ark., for plaintiff-intervenor Veo Easley.
Before ARNOLD, Circuit Judge, and OVERTON and WOODS, District Judges.
ARNOLD, Circuit Judge.
The issue before the Court is the constitutionality of Act 965 of 1981, Arkansas's congressional reapportionment plan for the decade of the 1980's. The total percentage variance among the populations of the four congressional districts under this statute is 1.87%, and the total population variance in absolute numbers is 10,667. The General Assembly had before it two other plans with a substantially smaller variance 0.78% and 0.75% respectively. Because the variance in the plan that became law was not unavoidable, and because it has not been justified in any legally acceptable way, the relevant precedents of the Supreme Court of the United States require us to declare the plan invalid under Article I, Section 2 of the Constitution, and we so hold.
Following the 1980 Decennial Census, the Census Bureau advised the State of Arkansas that its population was 2,285,513, an increase of 362,218 persons from the 1970 census. The increase did not occur equally in each of the four congressional districts, so it was necessary for the General Assembly to re-draw the district lines to some extent. Under the census total as it existed when the General Assembly met early in 1981, each district would ideally contain 571,378 people.
The Legislature had three principal proposals before it during its 1981 regular session. One plan was H.B. 883, whose principal sponsor was Representative (and former Speaker) Ray S. Smith, Jr., of Hot Springs. This plan was also introduced in the Senate as S.B. 622, by Senator Joe Ray of Havana, and it will be referred to as the Ray Plan. This bill passed the Senate on March 11, 1981, by a vote of 18 to 2, 18 being the least number of votes that a bill can receive in the 35-member Senate and still pass. Fifteen Senators were absent or did not vote. Under the Ray Plan, which never came to a vote in the House of Representatives, the maximum population variance would have been 0.78%. A map of the State as it would have looked under the Ray Plan is attached to this opinion as Exhibit A.
On February 23, 1981, the day before the Ray Plan was introduced in the House, Representative (and former Speaker) John E. Miller of Melbourne introduced H.B. 848, which we shall refer to as the original, or unamended, Miller Plan. Under this bill the maximum population variance would have been 0.75%. This plan was before the Senate as S.B. 600, introduced by Senator John Bearden of Leachville. On March 11, 1981 (the same day as it passed the Ray Plan), the Senate passed S.B. 600 by a vote of 20 to 8, with 7 Senators absent or not voting. The original Miller Plan, like the Ray Plan, never came to a vote on the floor of the House. Appended as Exhibit B is a map of the State as it would have looked under the original Miller Plan.
In the meantime, the House Committee on State Agencies and Governmental Affairs met to consider all four bills H.B. 848 and 883, and S.B. 600 and 622 as passed by the Senate and transmitted to the House. It soon became apparent that the Ray Plan did not have substantial support in the committee. Attention focused on the Miller bill, H.B. 848, and various amendments were offered in committee. The major amendment was sponsored by Representative Lloyd George of Dardanelle. The original Miller bill had placed Grant County in the Fourth District, Yell County in the Second, and Howard County in the Third. The George amendment proposed to move Grant County to the Second District, Yell County to the Third, and Howard County to the Fourth. Representative George's motive was to keep his home county, Yell, in the Third District, where it had been during the 1970's. His constituents liked their Congressman and wanted to keep him. The State Agencies Committee voted in favor of the George amendment, and reported out the Miller bill, H.B. 848, with the recommendation that it "do pass as amended" by the George amendment.
*1326 On March 12, 1981, the House agreed to the George amendment by voice vote. On March 16, 1981, after floor debate, H.B. 848 was passed by the House as thus amended. The vote was 76 to 13, with 4 present and 6 not voting. The principal opponent was Representative Smith, the House sponsor of the rival Ray Plan. His primary aim was to keep his home county, Garland, in the Third District, where it also had been during the 1970's. (Representative George had originally been a co-sponsor of the Ray Plan, but understandably abandoned it when he saw it would not pass the House.) In Representative Smith's view, his constituents (like Mr. George's in Yell County) liked their Congressman and wanted to keep him. The problem was that the Third District had grown faster than the rest of the state and so had to lose some territory.
After some further action in both houses not material here, H.B. 848 as amended received final passage and went to the Governor. He neither approved nor disapproved the bill, and on April 8, 1981, 20 days after adjournment of the Legislature sine die, it became law as Act 965 without the Governor's signature. As we have stated, the maximum population variance in the districts created by the Act is 1.87%. This increase over the variance of 0.75% in the Miller Plan as originally introduced was due to the shift of three counties under the George amendment. Appended as Exhibit C is a map of the State as redistricted by Act 965, which is now codified as Ark.Stat. Ann. §§ 3-401-05.
Also before us is a fourth plan, the so-called "Plaintiffs' Plan," which was not before the General Assembly, but which plaintiffs now ask us to order into effect, if we should hold Act 965 unconstitutional. Under the census figures as they existed when the Legislature was in session, this plan would have produced a total population variance of 0.46%. A map of the State as it would look under the Plaintiffs' Plan is appended as Exhibit D. Plaintiffs originally wanted this Court to order the use of the Ray Plan, but this request has now been abandoned.
On November 17, 1981, new census figures were received by the State. The population of eight counties was changed (seven increases and one decrease), and the total population of the State was slightly increased. Because of the distribution of the changes among certain counties, these new data cause the total percentage variances of the four plans to be somewhat different from what they appeared to be at the time of the legislative session. The changes may be summarized as follows:
Total Percentage Variance Under Original Under Revised Census Figures Census Figures Ray Plan 0.78% 1.17% Original Miller Plan 0.75% 0.78% Act 965 1.87% 2.10% Plaintiffs' Plan 0.46% 0.17%
We now turn to an examination of the case law that governs our disposition of the plaintiffs' claim. In order to put the issue in proper perspective, we repeat that there is no claim here that the Legislature was actuated by partisan considerations, or by a desire to put any group of citizens or Member of Congress at a disadvantage. In addition, there is not, nor could there be on the record before us, any suggestion that Act 965 has the purpose or effect of diluting minority voting strength. We have in mind also that none of the districting plans, neither those considered by the Legislature nor that now advocated by the plaintiffs, would require the crossing of county lines.
The constitutionality of the Act must be measured against the mandate of Article I, § 2 of the Constitution of the United States: "The House of Representatives shall be composed of members chosen every second year by the people of the several states ...." Wesberry v. Sanders, 376 U.S. 1, 7-8, 84 S. Ct. 526, 529, 11 L. Ed. 2d 481 (1964), interpreted this Section to mean that congressional districts should be drawn so that "as nearly as is practicable one man's vote in a congressional election is to be worth as much as another's." It further noted that "[w]hile it may not be possible to draw congressional districts with mathematical precision, that is no excuse for ignoring our Constitution's plain objective of making equal representation for equal numbers of people the fundamental goal for the House of Representatives." Id. at 18, 84 S. Ct. at 535.
Since Wesberry, the Supreme Court has forcefully reiterated that when state legislatures draw up congressional reapportionment plans, they must make a good-faith effort to create districts with equal populations. All avoidable deviations must be justified. The Supreme Court has explicitly rejected the notion that there exists a fixed de minimis population variance which need not be justified. White v. Weiser, 412 U.S. 783, 93 S. Ct. 2348, 37 L. Ed. 2d 335 (1973); Wells v. Rockefeller, 394 U.S. 542, 89 S. Ct. 1234, 22 L. Ed. 2d 535 (1969); Kirkpatrick v. Preisler, 394 U.S. 526, 89 S. Ct. 1225, 22 L. Ed. 2d 519 (1969).
Kirkpatrick v. Preisler, supra, held that a Missouri congressional reapportionment plan with a deviation of 5.97% or 25,802 persons, violated Article I, § 2. Missouri's main argument was that the population variances were de minimis and should therefore not require justification. The Court rejected this argument in absolute terms.
We reject Missouri's argument that there is a fixed numerical or percentage population variance small enough to be considered de minimis and to satisfy without question the `as nearly as practicable' standard .... [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 congressional districts are shown to have resulted despite such effort, the State must justify each variance, no matter how small.
Id. at 530-31, 89 S. Ct. at 1228 (citations omitted). The Court also disapproved the State's argument that the deviation was permissible because it represented a reasonable legislative compromise. "Problems created by partisan politics cannot justify an apportionment which does not otherwise pass constitutional muster." Id. at 533, 89 S. Ct. at 1230. The Court also stated that "`[n]either history alone, nor economic or other sorts of group interests, are permissible factors in attempting to justify disparities from population-based representation.'" Id., quoting Reynolds v. Sims, 377 U.S. 533, 579-80, 84 S. Ct. 1362, 1390, 12 L. Ed. 2d 506 (1964). Recognizing that congressional reapportionment plans usually remain in effect for ten years, the Court suggested that there might be no constitutional violation if the State could prove that the present population disparities among districts resulted from the legislature's effort to take into account projected population shifts.
Four years later, the Court addressed the constitutionality of Texas's congressional reapportionment plan in White v. Weiser, 412 U.S. 783, 93 S. Ct. 2348, 37 L. Ed. 2d 335 (1973). The Texas plan allowed a maximum deviation of 4.13% or 19,275 persons. Texas asked the Court to modify Kirkpatrick "so as not to require the `small' population variances among congressional districts involved in this case to be justified by the State." Id. at 792, 93 S. Ct. at 2353. The Court refused, id. at 793, 93 S.Ct. at 2353:
Keeping in mind that congressional districts are not so intertwined and freighted with strictly local interests as are state legislative districts and that, as compared with the latter, they are relatively enormous, with each percentage point of variation *1328 representing almost 5,000 people, we are not inclined to disturb Kirkpatrick and Wells. This is particularly so in light of Mahan v. Howell, 410 U.S. 315 [93 S. Ct. 979, 35 L. Ed. 2d 320] (1973), decided earlier this Term, where we reiterated that the Wesberry, Kirkpatrick, and Wells line of cases would continue to govern congressional reapportionments, although holding that the rigor of the rule of those cases was inappropriate for state reapportionments challenged under the Equal Protection Clause of the Fourteenth Amendment.
The Court affirmed the District Court's holding that the state plan was unconstitutional.
The District Court also had before it two alternative reapportionment plans, Plan B, allowing a maximum population disparity among the districts of .149% or 696 persons, and Plan C, with a deviation of .284% or 1,318 persons. The District Court adopted the latter, but the Supreme Court reversed. The Supreme Court based its decision on the grounds that Plan B contained a smaller deviation and followed the state's redistricting policy, while Plan C did not. Id. at 796, 93 S.Ct. at 2355:
Plan B, as all parties concede, represented an attempt to adhere to the districting preferences of the state legislature while eliminating population variances. Indeed, Plan B achieved the goal of population equality to a greater extent than did Plan C. Despite the existence of Plan B, the District Court ordered implementation of Plan C, which, as conceded by all parties, ignored legislative districting policy and constructed districts solely on the basis of population considerations. The District Court erred in this choice. Given the alternatives, the court should not have imposed Plan C, with its very different political impact, on the State. It should have implemented Plan B, which most clearly approximated the reapportionment plan of the state legislature, while satisfying constitutional requirements.
Our task is to apply the exacting standard of Kirkpatrick v. Preisler, as reaffirmed in White v. Weiser in an opinion in the relevant part of which all members of the Supreme Court joined. Because censuses occur only every ten years, there is no Supreme Court authority more recent than 1973, but it is our duty, nonetheless, to decide the case by the light that we have. As already stated,
the State [must] make a good-faith effort to achieve precise mathematical equality.... Unless population variances among congressional districts are shown to have resulted despite such effort, the State must justify each variance, no matter how small .... [The Constitution] permits only the limited population variances which are unavoidable despite a good-faith effort to achieve absolute equality, or for which justification is shown.
Kirkpatrick v. Preisler, supra, 394 U.S. at 530-31, 89 S. Ct. at 1228.
To begin with, it is clear on the face of it that the 1.87% variance contained in Act 965 was not unavoidable. The General Assembly had before it two other plans, the Ray Plan and the original Miller Plan, with variances less than half as large, 0.78% and 0.75%. (We use for this purpose the percentage variances produced by the original census figures, not those that would obtain under the revised numbers received from the Bureau of the Census in November of 1981. The Legislature cannot be expected *1329 to predict the vagaries of future census revision. The relationships among the variances produced by the three plans considered by the General Assembly are roughly the same under the new census figures, anyway, at least in the sense that Act 965, the plan that has become law, still has much the greatest variance.) In addition, it is true here, as it was in Kirkpatrick, that "resort to the simple device of transferring entire political subdivisions of known populations between contiguous districts would have produced districts much closer to numerical equality." 394 U.S. at 532, 89 S. Ct. at 1229.
The State insists that the strictures of Kirkpatrick can be avoided because here the General Assembly did act in "good faith." The argument is not without appeal. As an original matter, one might have thought that a variance of less than two per cent could be tolerated, at least where, as here, the Legislature has acted with no invidious motives. We have no doubt that the Arkansas General Assembly conducted itself in entire good faith, from motives honestly political, in the sense that it strove to reach a solution acceptable to most of the constituencies and interest groups involved, while still hewing fairly closely to the line of mathematical equality. In addition, the Legislature of Arkansas has no history of recalcitrance, nor did it give the question of reapportionment merely cursory or haphazard consideration, as Missouri seems to have done in Kirkpatrick.
In her closing argument, able counsel for the State urged that these factual differences from Kirkpatrick, coupled with the fact that the population variance here is less than one-third what it was there, should produce a different result in the instant case. Certainly if the question were governed by the somewhat less exacting standards of the Equal Protection Clause of the Fourteenth Amendment, we should have little trouble in agreeing. We think, however, that the result must be otherwise under Article I, § 2. The Supreme Court has indicated that a variance of two per cent is unsatisfactory. The Kirkpatrick opinion mentions with apparent disapproval the notion of "one leading Missouri legislator" that it was "proper to attempt to achieve a 2% level of variance rather than to seek population equality." 394 U.S. at 531, 89 S. Ct. at 1229. The very distinguishing factors the State urges here, including the Missouri Legislature's apparent unwillingness to conform to federal standards, are stressed in the concurring opinion of Mr. Justice Fortas in Kirkpatrick, 394 U.S. at 536, 540-41, 89 S. Ct. at 1231, 1233-1234. He would have struck down the Missouri plan on grounds narrow enough to permit us to uphold the Arkansas law at issue here. This view was before the Supreme Court and was necessarily rejected by its use of more sweeping language in its Kirkpatrick opinion. We conclude that the use of the term "good faith" in Kirkpatrick cannot mean that every State plan that involves no invidious motives and comes reasonably close to equality is to be upheld. Rather, when population variances can be greatly reduced simply by moving counties of known population between contiguous districts, the disparity is unavoidable, and there can be no "good faith" in the special sense in which Kirkpatrick uses that phrase. We repeat that the Legislature here in no sense acted in bad faith in any commonly accepted sense of that term. The same thing could have been said in Kirkpatrick, where there was "no finding of gerrymandering." 394 U.S. at 541, 89 S. Ct. at 1234 (Fortas, J., concurring).
We think it significant, in addition, that, although 1.87% is a relatively small proportion, it represents over 10,000 people in absolute numbers, hardly a negligible group. Under Act 965, residents of the Second District are given relatively more *1330 voice in the choice of a Member of Congress than are residents of the other three districts, because only the Second District has a population below the average of all four. The dilution, in any practical sense, is hardly great, but the Supreme Court has held that in this field there is no such thing as de minimis.
The State also suggests, in its pre-trial brief, that the variance is justified by similarity of economic interests, by history, by accessibility of voters to their Member of Congress, and by the necessity of reaching some solution that would command a majority of both houses of the General Assembly. Each of these justifications was considered and rejected in Kirkpatrick, prompting Mr. Justice Fortas to complain that the Court rejected "every type of justification that has been possibly, every one that could be advanced." 394 U.S. at 537, 89 S. Ct. at 1232 (concurring opinion). Beyond that, there is simply no basis here for a finding that these factors explain, in any systematic way, the General Assembly's choice of Act 965 over the unamended Miller bill. Representative Miller, the State's principal witness, all of whose testimony we credit, said as much from the witness stand. He explained that timber, tourism, and many other comparable considerations were mentioned in debate by both sides, but he readily granted that neither timber interests nor tourism could justify the variance in Act 965. He preferred, of course, his original bill, and accepted Act 965 only because it was as close as he could come given the practicalities of obtaining a majority in the House of Representatives.
Representative George gave perhaps the best description of the process in his deposition, p. 51:
Well, this is all political, yes. Yes, I'm saying that. I'm saying that I regret the population numbers can't be exactly zero just no variation whatsoever. I'm firmly convinced that the plan we've passed is as close as possible as we can get to the numbers and pass the bill ....
We are not among those who think "politics" is a dirty word. Politics, in its higher and better sense, is the life blood of our form of government. But the Supreme Court has explicitly rejected political practicality as a justification for population variance. "We also reject Missouri's argument that `[t]he reasonableness of the population differences in the congressional districts under review must ... be viewed in the context of legislative interplay. The legislative leaders all testified that the act in question was in their opinion a reasonable legislative compromise.... It must be remembered ... that practical political problems are inherent in the enactment of congressional reapportionment legislation.' We agree with the District Court that `the rule is one of "practicability" rather than political "practicality."' [Preisler v. Secretary of State of Missouri], 279 F.Supp.  at 989." Kirkpatrick v. Preisler, supra, 394 U.S. at 533, 89 S. Ct. at 1230.
The State's case also included testimony about projected population changes, and such considerations may be, in some circumstances, a justification that the Supreme Court would accept. If, for example, it could be demonstrated that one district will grow faster than the other three between now and 1990, it might be permissible to assign fewer people to that district initially. This argument is unavailing here for several reasons. First, although Representative Miller testified that he took such considerations into account in preparing his original bill, he also testified that projected population growth was more of an argument for his original plan than for Act 965. The latter plan, among other things, places more people in the new Third District than in any other, and according to Mr. Miller the Second and Third Districts should be the two fastest-growing areas in the next ten years. We cannot say, in any case, that population shifts "can be predicted with a high degree of accuracy," as required by the Supreme Court in Kirkpatrick, 394 U.S. at 535, 89 S. Ct. at 1231, as a predicate for this kind of justification, nor that "[f]indings as *1331 to population trends," in this case, have been "thoroughly documented and applied throughout the State in a systematic, not an ad hoc, manner." Ibid.
It has been suggested that White v. Weiser, supra, may have relaxed the strictures of Kirkpatrick. We disagree. There, a plan including a variance of 4.13% in total population was struck down. The State urged that the variance was small and that there was no proof of invidiousness. The Court replied that the percentage deviations were not "`unavoidable,' and [that] the districts were not as mathematically equal as reasonably possible. Both Plans B [an alternative with a .149% variance] and C [with a variance of .284%] demonstrate this much ...." 412 U.S. at 790, 93 S. Ct. at 2352. The Court also noted the passage by one house of the Texas Legislature of a bill with lower deviations on two occasions. "Although both bills were ultimately defeated in the Senate, their passage by the House, and indeed their very existence, indicates that it was possible and practicable to construct a redistricting scheme with lower population deviations among districts than those embodied in" the bill that passed. Id. at 790 n.9, 93 S. Ct. at 2352, n.9. So here, the passage by the Arkansas Senate of the Ray Plan, and the original Miller bill, indicates that it was possible and practicable to improve upon Act 965. The White opinion also states that "a district court should ... honor state policies in the context of congressional reapportionment," id. at 795, 93 S. Ct. at 2354, but this language occurs in Part II of the opinion, and is limited to the remedial phase of the case. State policies are relevant at this stage, when a court may have to fashion a reapportionment plan of its own or choose among plans suggested by the parties. They are not relevant (except in those rare instances, up to now unrevealed, in which a state policy may be a legal justification for population variances) to the initial decision whether the State's own plan is constitutional. State policies are to be considered in the remedial phase of a case, moreover, only when "adherence to state policy does not detract from the requirements of the Federal Constitution." White, 412 U.S. at 795, 93 S. Ct. at 2354. See also id. at 797, 93 S. Ct. at 2355. We conclude that Kirkpatrick retains its full vigor. As recently as 1975, a unanimous Supreme Court, citing Kirkpatrick and White, observed that in congressional districting "population equality appears now to be the preeminent, if not the sole, criterion on which to adjudge constitutionality ...." Chapman v. Meier, 420 U.S. 1, 23, 95 S. Ct. 751, 764, 42 L. Ed. 2d 766 (1975).
Defendants cite two post-Kirkpatrick cases in which congressional districting plans were upheld and suggest that they support a judgment upholding Act 965. We disagree. In West Virginia Civil Liberties Union v. Rockefeller, 336 F. Supp. 395 (S.D. W.Va.1972) (three-judge court), the plan upheld contained a variance of only 0.78%, exactly the variance under the Ray Plan at the time it passed the Senate, and under the original Miller bill when judged by the revised census figures. The Legislature had considered 17 plans, none of them involving a total variance of more than one per cent. The Legislature had employed a statistician and sociologist as a consultant and had used a computer to produce and evaluate redistricting proposals. Plaintiffs offered no plan, nor had one been considered by the Legislature, that would have produced a variance markedly reduced from 0.78%. Rather, other plans available would have provided "only slight reductions in the variances," 336 F. Supp. at 400. Here, the Legislature actually considered and rejected two plans that would have reduced the variance by more than one per cent.
Our attention is called also to Drum v. Scott, 337 F. Supp. 588 (M.D.N.C.1972) (three-judge court). There, a legislatively adopted plan with a total variance of 3.79% was upheld, despite the fact that the Legislature had considered and rejected a plan with less than half the average deviation *1332 from absolute equality. The Court emphasized the reference in Kirkpatrick to "good faith," and held that "Kirkpatrick and Wells curtail, but do not destroy, the `de minimis' concept." 337 F. Supp. at 590. The State's citation of Drum is well taken. We simply disagree with the case's reading of Kirkpatrick, for reasons already explained in our discussion of the Supreme Court's use of the term "good faith." In addition, we doubt that Drum can survive White v. Weiser, in which a 4.13% variance was condemned, and the fact that the Legislature had considered and rejected a more nearly equal plan was relied on virtually as proof positive that the variances occasioned by the plan adopted were not "unavoidable."
In sum, this case is governed by the opinions of the Supreme Court in Kirkpatrick and White. We have read, marked, learned, and inwardly digested them, and hold that they require a judgment declaring Act 965 unconstitutional. The Supreme Court is of course at liberty to reexamine those cases. As an inferior court, we are not.
It remains to consider what remedy, in addition to the declaratory judgment already mentioned, is appropriate on equitable principles. Plaintiffs ask us to order their new plan (see Appendix D to this opinion) into effect at once. We decline the invitation. The plaintiffs' plan was not considered by the General Assembly, or even conceived of at the time of the regular session. It would have a "very different political impact," White v. Weiser, supra, 412 U.S. at 796, 93 S. Ct. at 2355, from Act 965. For example, the plaintiffs' plan would place Desha County, in the Delta country of Southeast Arkansas, in the same congressional district with the North Arkansas mountain counties of Newton and Searcy, a result that must inevitably raise the eyebrows of anyone conversant with Arkansas history and politics. It is also not so compact as the Ray Plan and the original Miller bill, for whatever that may be worth. The General Assembly could of course adopt the plaintiffs' plan if it wishes, but we will not do so, at least in the first instance. The Legislature should be given a chance to adopt a new plan. "[S]tate legislatures have `primary jurisdiction' over legislative reapportionment." White, 412 U.S. at 795, 93 S. Ct. at 2354.
The State asks that we stay the effective date of the holding that Act 965 is invalid and permit the 1982 congressional elections to be conducted under the existing law. The 1983 regular session (the next one scheduled) of the General Assembly would then consider a new plan in due course. This suggestion would save the expense of an extraordinary session of the Legislature. Counsel for the defendants advised us in oral argument that the last special session, held in 1981, cost $491,000. That session lasted, we should think, considerably longer and therefore cost considerably more than would a session limited to the consideration of a new congressional-districting bill. In any event, cost cannot be a controlling consideration where constitutional rights are at stake. Cf. Jackson v. Bishop, 404 F.2d 571, 580 (8th Cir. 1968) (Blackmun, J.) ("constitutional requirements are not, in this day, to be measured or limited by dollar considerations ....") (application of Eighth Amendment to state prison practices cannot be limited by monetary considerations). The filing period for candidates for Congress opens on March 16, 1982, and closes on March 29. The primary election is scheduled for May 25, with a run-off on June 8. Ark.Stat.Ann. § 3-113(a), (c), (d). The general election, of course, does not take place until November. There is ample time for the Legislature to meet and adopt a new plan if it and the Governor wish.
We therefore withhold further action in order to give the General Assembly a chance to work its will within constitutional limits. The question could fairly be asked and no doubt will be what kind of *1333 plan would be valid, if Act 965 is not. We think either the Ray Plan or the unamended Miller bill would comply with Article I, § 2, although the original Miller bill involves a somewhat smaller variance, particularly under the revised census figures, which the General Assembly would have to take into account if the Governor decides to call it into session again. Counsel for plaintiffs conceded at argument that the original Miller bill would be valid, and the Ray Plan was plaintiffs' original preference.
It seems to us that February 15 is a reasonable date by which the Legislature may be expected to have acted, if it chooses to act at all. Failing the adoption of a new congressional-districting law by that time, it will be our duty to consider further relief. We could leave the State without districts, in which event all four congressional seats would be filled by elections at large throughout the State. That was the court's view in Park v. Faubus, 238 F. Supp. 62, 65-66 (E.D.Ark.1965) (three-judge court), and it has much to commend it. Neither the Constitution nor any Act of Congress requires the States to divide themselves into congressional districts. The last federal statute requiring that States so divide themselves, former Section 3 of 2 U.S.C., expired by its own terms in 1929. Arkansas has always had congressional districts, however, as far as we know, and for four members of Congress to have to run at large with barely more than two months' notice would indeed be a radical departure from the practice and policy of this State. The Park court acknowledged that "[t]he better practice may be to divide the State into districts," 238 F. Supp. at 66, and we will do so here if necessary. That appears to be the course adopted by most district courts that have had to face the issue. See, e.g., In re Illinois Congressional Districts Reapportionment Cases, No. 81-C-3915 (N.D.Ill.1981) (three-judge court); LaComb v. Growe, No. 4-81-414 (D.Minn. Sept. 14, 1981) (three-judge court) (in the absence of action by the Minnesota Legislature, the court will put into effect a plan with no more than 0.50% total population variance). It is of some moment that no party to this case advocates at-large elections. If we must order a plan into effect, it is our present intention to choose either the Ray Plan or the original Miller bill, subject, of course, to whatever suggestions or arguments the parties may make to us in a timely fashion. In making this choice, if it comes to that, we will have in mind, among any other factors that may be relevant, that the original Miller bill has a smaller population variance than the Ray Plan, got more votes in the Arkansas Senate, and seems closer to the design of Act 965.
Judgment will be entered declaring Act 965 unconstitutional and enjoining the defendants from conducting a congressional election in accordance with its terms. If the General Assembly of Arkansas does not enact a valid new plan by February 15, 1982, we will reluctantly proceed to put in force a new division of the State into congressional districts, to be effective beginning with this year's elections.
IT IS SO ORDERED this 5th day of January, 1982.
*1334 APPENDIX A
*1335 APPENDIX B
*1336 APPENDIX C
PLAINTIFFS' PROPOSED PLAN Numerical Percentage District Population Variance Variance District 1 572,047 +438 +0.08 District 2 571,873 +264 +0.05 District 3 571,451 -158 -0.03 District 4 571,064 -545 -0.09 Total 2,286,435 ±0.17
*1337 APPENDIX D NOTES
 After Act 965 was passed, revised census figures were received, giving a new population of 2,286,435. These new figures cause the variances in Act 965 and the various other plans discussed to be somewhat different from what they were when the Legislature met, as we will indicate in more detail below.
 The four districts would have had the following populations if this plan had been adopted:
Percentage Population Variance First District 569,494 -0.33% Second District 571,593 +0.04% Third District 574,010 +0.45% Fourth District 570,416 -0.17%
 The four districts would have had the following populations if the original Miller bill had been adopted:
Percentage Population Variance First District 573,128 +0.31% Second District 568,836 -0.44% Third District 571,918 +0.09% Fourth District 571,631 +0.04%
 Many of the cases we have read in our attempt to divine the law on this subject involve partisan conflict, an attempt by legislators of one party to disadvantage the other. No such claim has been made here. One of the ironies of this case is that a good deal of the disagreement among House members in passing a reapportionment bill derived from their wish to keep their various constituencies within the district of a Republican Congressman. The Arkansas House of Representatives is overwhelmingly Democratic.
 Under Act 965 the four congressional districts have the following populations:
Percentage Population Variance First District 573,128 +0.31% Second District 564,818 -1.15% Third District 575,485 +0.72% Fourth District 572,082 +0.12%
 The distinction between congressional-district cases and state-legislative-district cases had been made inescapably clear in Mahan v. Howell, 410 U.S. 315, 322, 93 S. Ct. 979, 984, 35 L. Ed. 2d 320 (1973), where the Court said (emphasis added):
Thus, whereas population alone has been the sole criterion of constitutionality in congressional redistricting under Art. I, § 2, broader latitude has been afforded the States under the Equal Protection Clause in state legislative redistricting ....
Accord, Gaffney v. Cummings, 412 U.S. 735, 741-42, 93 S. Ct. 2321, 2325, 37 L. Ed. 2d 298 (1973).
 The 1971 reapportionment produced congressional districts with a total population variance of 0.28%.
 Except, of course, when we had only one Representative in Congress, which was apparently the case from 1836 to 1853. See Park v. Faubus, 238 F. Supp. at 66 n.6.
 We are advised that an appeal to the Supreme Court is now pending in the Illinois case.