ALTERNATIVE GOALSFAIR CRITERIA AND PROCEDURES FOR ESTABLISHINGLEGISLATIVE DISTRICTS*Robert G. Dixon, Jr. , Washington University, St. Louis

When the districting of legislative seats was opened up tojudicial scrutiny in 1962 in Baker v. Carr (1962)^, under thefederal Constitution's equal protection of the laws principle, thussignaling the beginning of the "one man-one vote" revolution.Justice Felix Frankfurter made the following observation on theguidance the Supreme Court in Baker gave to the lower courts:

Room continues to be allowed for weighting. This ofcourse implies that geography, economics, urban-ruralconflict, and all the other non-legal factors which havethroughout our history entered into political districtingare to some extent not to be ruled out in the undefinedvista now opened up by review in the federal courts ofstate reapportionments. To some extent — aye, there'sthe rub.

—Baker v. Carr (Frankfurter dissenting)^

The extent to which anything other than equality of district popu-lation can be or must be taken into account in political districtingis still the "rub." We approach the next round of reshaping leg-islative districts—national, state, and local—which will be triggeredby the 1980 census without clear judicial guidance on constitutionalground rules or an informed consensus on proper method.

Indeed, the "rub" is even worse than this Frankfurtercomment signaled. The first key fact is that whether or not non-population factors are expressly taken into account in shapingpolitical districts, they are inevitably ever-present and operative.They influence all election outcomes in ail sets of districts. Thekey concept to grasp is that there are no "neutral" lines forlegislative districts. Whether the lines are drawn by a ninth-gradecivics class, a board of Ph.D.'s, or a computer, every line drawnaligns partisans and interest blocs in a particular way differentfrom the alignment resulting from putting the line in some otherplace. And the electoral result will be different, bearing in mindthat the gross majority in each district captures the seat or seatsassigned to that district.

A second key fact is that at any level of equal populationstringency — 10 percent maximum deviation of a district from

*This paper draws upon Dixon (1968, 1969, 1971) and Dixonand Hatheway (1969). It was initially presented as testimony onS. 596, "A Bill to Provide a Fair Procedure for Establishing Con-gressional Districts," before the Committee on Governmental Affairs,United States Senate, June 20, 1979. After Professor Dixon'sdeath, with the consent of his wife, Claire Dixon, this testimonywas edited fcr inclusion in this volume.

839

ideal district, or 5 percent, or 2 percent, or one percent, orless — a computer can churn out not just one but hundreds ofequally "equal" districting plans. But each plan, because of itssomewhat different grouping of partisans and interests, will havea different — and of course nonneutral — impact on the electoraloutcomes. We must consider electoral outcomes in two senses.First, there is the outcome in a given district, where the winner-take-all principle operates. Second, there is the outcome for thelegislature as a whole where an important part of the power equa-tion, and the reality of effective representation, depends on cap-turing enough district seats to win a iegisiative majority and beable to control the key committee chairmanships.

A third key fact is that even with equal population stringen-cy in the drawing of district lines, the manner in which the"winner-take-all" majorities and minorities in each district add upacross all the districts in a state can produce gross inequalities,even a minority election. In Connecticut in 1970 under a one man-one vote plan the Republicans polled a statewide popular majorityin the legislative seat voting for both houses, yet captured amajority of seats in neither house. In Iowa in 1966, although itwas not a minority election, the Republicans captured the con-gressional delegation, 5 - 2 , with only a hairline majority of thestatewide vote polled for congressional seats. In California in1966, in the first election under an equal population revision of thedistrict lines, the Democrats won a majority of the seats in eachhouse of the state legislature, although the Republicans polled astatewide majority of the legislative vote for each house.

A fourth key fact, and the saddest of all because 17 yearshave passed since Baker v. Carr, is that the first three key factsare not understood by judges who rule on these matters, manyjournalists who report these matters, and many members of thegeneral public. There is a sort of vague impression in manyquarters that equality in census numbers alone produces basicfairness, that legislative district lines can be politically neutral,that something called nonpartisanship can be built into the dis-tricting process. My own experience tells me that although I mayfind nonpartisanship in heaven, in the real world, and especiallyin academia, there are no nonpartisans, although there may benoncombatants.

GOALS AND DIFFICULTIES IN LEGISLATIVE DISTRICTING. Thecore problem can be stated somewhat simply: our ideals aboutpolitical representation and our implementing election system do notfit together neatly. One of the major ideals, here as well as inEurope, is to have the political parties, who after all still organizeand run our legislatures, win seats in legislatures roughly pro-portional to their share of the popular vote. That is the coremeaning of the term "fair representation." To the extent thatother interests can be factored in, we probably would like thatalso. Such an effort would be a highly speculative process, how-ever, and probably cannot be carried beyond recognition of thefact that the support for any candidate is itself a collection of"interest factions," with varying degrees of internal organization,

840

cohesiveness, and visibility. A key stumbling block to formalattempts to factor in subinterests is that district representationis a zero-sum game. Explicit favoring of one subgroup adverselyaffects another group (and the political party with which it isaligned). That is the lesson of United Jewish Organizations v.Carey (1977), ̂ and was a background issue in Whitcomb v. Chavis(1971).4

This fair representation ideal, which might be called the idealof "proportionate representation" of parties (and the interests sub-sumed within them), does not dovetail well with an eiection systembased on use of geographic legislative districts and the pluralityrule (or winner-take-all rule) within each district. Obviously, ifthere were 10 congressional districts in a state and the Democraticparty polled 55 percent of the vote in each district while theRepublicans were polling 45 percent in each district, the Democratswould have a disproportionate 100 percent of the representation inthe congressional delegation and the Republicans would have a dis-proportionate zero percent of the representation. Indeed, as iswell recognized in poiiticai science, even under ideal circumstancesa district system of election Iegisiative representatives always tendsto overrepresent the dominant party in a given eiection year.That party's dominance tends to be reflected across many dis-tricts — certainly across all of the so-called balanced or "swing"districts.

But we live with this system because we are used to it, andbecause it accomplishes certain other ideals, or at least beneficialresults. It accomplishes the ideal (or strongly tends to) of pre-serving a two-party system. A two-party system operates to pro-duce such coordinate goals as a clear governing majority, govern-mental stability, and pin-pointing of governing responsibility. TheEuropeans do not use our kind of district system but use propor-tional representation systems whereby parties are guaranteed seatsin proportion to their percent of the popular vote. However, aproportional representation system strongly tends to invite theformation of a multiparty system; even a small party can get somekind of a "win" in terms of seats in parliament. The result usuallyis that no party wins a majority of the seats, thus necessitatinggovernment by unstable coalitions of minority parties. In short,proportional representation election systems yield more propKjrtion-ate representation than do district systems, but sacrifice thecoordinate goals of a governing majority, governmental stability,and clear lines of responsibility.

There is no possibility, of course, nor do I now recommendit, of our moving from our present district system of electinglegislators, with its combination of stability and inexact represen-tation, to a formal system of proportional representation with itsmore precise representation of ideology and emotion but greaterinstability. Nevertheless, to appreciate what we have — itsvirtues and its problems — it is helpful to note in passing thisalternative system which dominates Europe and perhaps is moreeasily copied than our own. It substantially avoids the verydifficult process of districting, avoids gerrymandering, and pro-duces significant fairness in representation.

841

THE BIPARTISAN COMMISSION PROPOSAL. There is a possibility,however, of improving our procedure for redistricting. Our ex-perience with almost two decades of the reapportionment and redis-tricting revolution shows that the device of bipartisan commissionwith tie breaker not only has logical appeal but is better than anyother available device. It also can work in practice, as in Connect-icut in 1971. [See Caffney v. Cummings (1973).]^ The bipartisancommission with tie breaker device allows combining the populationequality principle with poiiticai realities and a better informed pub-lic scrutiny.

The bipartisanship is an essential built-in check on bothconscious and unconscious unfairness in the resulting districts.Although it is never possible to guarantee "fairness" in a districtsystem of election (any more than the Federal Trade Commissioncan guarantee fairness in competition), it is possible for a bipar-tisan commission to discard plans that are predictably unfair in thelight of all that is known about the political behavior of the areain question. Indeed, this mode is the way the FTC was designedto operate — to negate, case by case, "unfair methods of compe-tition." Obviously, to accomplish the "unfairness policing" mission,either of the FTC or of a bipartisan districting commission, a broadintake of all relevant data is necessary. For the FTC this meansdata bearing on all aspects of competition. For a bipartisan com-mission it means data bearing on all aspects of political and elector-al behavior, because the bipartisan commission simply deals withcompetition in another form — political competition for politicalseats in the poiiticai assembly.

The need to give balanced consideration to political data ifavoidance of disproportionate electoral outcomes is the goal inlegislative districting is the basis for a recent districting pro-cedure article by two political scientists. They write: "If weaccept the premise that seats in a representative body shouldchange in some specified way as vote totals change, then it isapparent that there is a need to incorpxjrate the partisan divisionof the vote into the criteria for fair districting" (Niemi and Deegan,1978: 1304).

Given the districting realities sketched above, and the po-tential for improvement offered by the bipartisan commission de-vice, it necessarily follows that care must be taken not to tie thehands of the bipartisan commission unduly in the course of speci-fying "standards" to guide their discretion. A bipartisan com-mission with blinders which can consider only census populationequality among districts would be as useless as a Federal TradeCommission which could look only at market price. The most im-portant injunction is that in its necessary consideration of data onelectoral behavior the redistricting body should do so in order totest and discard unfair pians, and not for the purpose of manu-facturing artificial majorities in the legislative assembly. The ruleshould be sameness or fairness of treatment to all parties, i.e.,"neutrality" in this special sense. The vice to be avoided isdifferential advantage, one party over another, in the cause of

842

manufacturing a congressional delegation majority out of a reason-ably predictable statewide minority of the popular vote.

If a commission were to be barred from use of the poiiticaiaffiliation of registered voters, previous election results, ordemographic information other than population head counts, thenthe bipartisan commission would have no rational function.^ In-deed, it would be demeaning to serve on such a commission. Thecommission would be making some very basic decisions about thepolitical future of the state—the make-up of its congressional dele-gation being the "prize" at issue—but would be doing so utterlyin the dark. The bipartisan commission should be free to con-sider voting behavior data for the purpose of avoiding a set ofdistricts which would favor one p>arty over another in the contestfor control of the legislature or state's congressional delegation.The congressional delegation, in states with several congressmen,offers opportunities for minority party representation that are notpossible in the statewide election of one United States Senator ata time on a staggered-term basis. I submit that it should be un-thinkable to pick as the final redistricting plan, from among themany "equally equal" plans available in population terms, the planthat predictably favors one p)arty over another at the instant ofenactment. Such a result should be equally abhorrent whether thebuilt-in favoritism was purposely planned, or the careless resultof consciously uninformed decision-ma king. Such an abhorrent re-sult can be avoided only by considering political data to test thedegree of predictable bias in the proffered plans.

THE CONSTITUTIONAL LAW OF REAPPORTIONMENT-DISTRICTING.

Historical Development. Until 1962 the courts followed a hands-offpolicy in respect to legislative districting. Justice Frankfurter inhis 1946 controlling opinion in Colegrove v. Creen, which involvedthe congressional districts in Illinois, had warned that it wouid be"hostile to a democratic system to involve the judiciary in thepolitics of the people."^ The problem before the Court then, as in1962 in Baker v. Carr when the Court decided to enter the "px)liti-cal thicket," was gross population malappxDrtionment. At the timeof Colegrove v. Creen the population disparity between the largestand smallest congressional districts in Illinois was the most extremein the nation: 914,053 to 112,116. At the time of Baker in 1962,which was a challenge to the malapportioned state legislative dis-tricts in Tennessee, disparities between largest and smallest dis-tricts of 10 to 1 or higher were common in most states for bothcongressional and state legislative districts. In Tennessee, forexample, lower house districts ranged from 42,298 down to 2,340.

Baker v. Carr was responsive to the fact that p>oliticalavenues for redress had become dead-end streets. With dis-parities of this dimension, any serious move toward equalizationwould destroy the districts of a substantial number of legislators.The ultimate rationale for Baker v. Carr was that some judicialintervention into the "politics of the people" had become necessaryto make democratic politics effective. When the Supreme Courtreached the merits two years later in 1964, it predictably

843

announced a constitutional principle of population equaiity. It saidin Wesberry v. Sanders (1964)^ that congressional districts shouldbe equal "as nearly as is practicable," and said in Reynolds v.Sims (1964)^ that state legislative apportionments must be "basedsubstantially on population." The difference in wording was in-consequential, and for a time the cases were cited interchangeably.

Within a short two or three years the problem that had givenrise to these cases had been corrected. That problem, to repeat,was gross population malapportionment due to legislative nonaction,which normally resulted in substantial overrrepresentation of ruraland small town areas and substantial underrepresentation of thegrowing urban and suburban areas. In other words, the problemwas one of regional imbalance in political control, in large measure.

However, the reapportionment-redistricting revolutionquickly transcended its origin. Because the concept of substantialpopulation equality is not a self-defining concept, there was astrong tendency in lower courts, with the Supreme Court following,to make the population equality requirement ever more stringent.

Two tactical factors contributed. First, the courts began toinsist that all population deviations be justified in terms of a con-sistent, logical application of identifiable state p>olicies. This wasa practical impossibility, both because all law-making, includingredistricting, is a compromise and adjustment process and not anexercise in logic, and more importantly, because nonpopulationpolicies — even a policy of following poiiticai subdivision linesinsofar as practicable — cannot be made objective. Second, it wasalways easy for a plaintiff to offer a slightly more "equal" planthan the official state plan (albeit also more palatable to theplaintiff's political interests which motivated his suit). Hence,there was pressure on the state to move in the direction of ever-tighter equality in order to maximize the chances of prevailingin court, but plaintiffs still frequently prevailed with last-minute"tighter" plans.

A direct corollary of insistence on ever-smaller p>opulationdeviations — from 15% which was an early rule of thumb down to10%, 5%, or lower — was to maximize the need to cut ever morepolitical subdivision lines. This also maximized the freedom ofchoice in drawing new lines, and consequently greatly increasedopportunities for politically imbalanced districting, i.e., gerry-mandering.

A literal sea of litigation resulted, and the issue likewiseshifted. The issue shifted from the relatively simple question ofsafeguarding against gross population disparities to the complexquestion of safeguarding against misrepresentation of interests.The latter problem arose as district lines were fine -tuned nomi-nally in pursuit of population equaiity but concurrently and pre-dominantly in pursuit of one or another political result.

Some sense of this problem was signaled murkily in ReynoldsV. Sims, but no solution — no reliable guidelines — were given.Consider the following caveats in Chief Justice Warren's opinion:

844

"We realize that it is a practical impossibility to arrangelegislative districts so that each one has an identicalnumber of residents or citizens or voters, "-^^

"[indiscriminate districting without any regard forpolitical sub-divisions or natural or historical lines,may be little more than an open invitation to partisangerrymandering. "^^

" . . . fair and effective representation for all citizensis concededly the basic aim of legislative apportion-ment. "^2

Especially significant is this last quote. It says that some-thing more than "equal numbers" is needed. It suggests in effectthat the "payoff" in terms of voice and influence inside legislatureis relevant, even predominant. Under this latter theme, what datadata is relevant? Is it necessarily essentially political data?

To a large extent we are still wallowing in the wake of theconfusions implicit in Reynolds v. Sims. At the time, Paul Freundobserved that the opinion reminded him of the little boy who hadjust learned how to spell the word "banana" — ba-na-na-na-na —but didn't know when to stop. Somewhat later Archibald Cox ob-served, in reference to several Warren Court initiatives under theequal protection of the laws principle, that "once loosed, the ideaof Equality is not easily cabined" (Cox, 1966: 91).

Current Population Equality Rules. As if in Pavlovian response tothis observation the Supreme Court in two congressional districtingcases in 1969 from Missouri and New York went all the way andmandated a virtual absolute equality rule for congressional dis-tricting. In his opinion for the Court Justice Brennan said theConstitution "requires a State to make a good-faith effort toachieve precise mathematical equality (Kirkpatrick v. Preisler) .-^-^He then proceeded to eviscerate every conceivable countervailing"fair representation" concept, including all the caveats of ChiefJustice Warren's opinion in Reynolds v. Sims. Four justices —Harlan, Stewart, White, and Fortas — in strong terms rejected themajority rationale and stressed the gerrymandering freedom undera rule allowing district lines to run anywhere. Indeed, the plain-tiff in the companion case from New York, Wells v. Rockefeller,^^asked: "Can we appeal from a decision that we won?" (Dixon,1969: 219, 226). The maximum deviations from ideal population inthe invalidated plans were 3.1% (Missouri) and 6.6% (New York).

Now it may be, to give Justice Brennan his due, that con-gressional districts are so large and amorphous that cutting in twoany given hamlet — or apartment house, as Justice Fortas hypoth-esized in bringing the new rule to its reductio ad absurdum — inpursuit of absolute population equality is immaterial. At least so Imight have thought until three or four years ago when I moved toWashington University and a home in Clayton, Missouri. Claytondirectly abuts the City of St. Louis and is the county seat of St.Louis County, which now dwarfs the City of St. Louis. Yet I findthat this community has been split and a small p>art of it is in a

845

congressional district dominated by an adjacent part of tiie City ofSt. Louis with quite different political representation concerns.Under any concept of "fair and effective representation" this doesnot make sense.

This "precise mathematical equality" rule of the 1969 con-gressional district rulings was reendorsed for congressional dis-tricts in 1973 in White v. Wei'ser. ^^ The anomaly of Weiser is thatit endorsed a rule which three members of the Court (the ChiefJustice and Justices Powell and Rehnquist) said they would nothave supported had they been on the bench at the time of Kfrk-patrick-Welfs, and that two other Justices (White, who wrote theopinion in Wefser, and Stewart) had opp>osed at its inception in1969. A further anomaly is that although it is true, as JusticeWhite observed, that a 1% deviation from ideal in congressionaldistricts averaging 450,000 is 4,500 census bodies, the plan theCourt favored in Weiser transcended census accuracy. There isan acknowledged margin of error in the census of 2%, which yields9,000 census bodies in the context of congressional districting.The districting plan the Court favored was claimed to have a maxi-mum deviation from ideal, in census terms, of 400 in excess of theideal and 10 under the ideal.

Concurrently in 1973, however, in state legislative reappor-tionment cases from Virginia and Connecticut — Mahan v. Howeli^^and Caffney v. Cummfngs — the Supreme Court clarified the popu-lation equality ground rules for state legislative districts. InMahan a plan in which the most deviant district was 9.6% fromideal passed muster because it allowed preservation of almost allpolitical subdivision lines. It may be an atypical case because fewstates have such an even spread of population as among the Vir-ginia counties. In Caffney, in more generally applicable terms forstate legislative districts, the Court endorsed something analogousto a rule that districts are prima facie constitutional in populationterms if the population deviations are de minimis in relation to theavailable census data. In this instance the deviations in terms ofactual census head counts were under 800, and the average popu-lation of the census units available to work with was 1,100.

Cerrymandering. Meanwhile, on the issue of fair districting interms of legislative payoff (gerrymandering issue) the SupremeCourt has shown great disinclination to get involved. The presentstate of the law seems to be that gerrymandering claims may bejusticiable, i.e, courts could consider them, but plaintiffs have theburden of proving invidious discrimination. The earlier cases in-volved multimember districts, which of course tend to submergeboth political and racial minorities under the winner-take-all rule.Fortsan v. Dorsey (1965) ; ^^ Burns v. Richardson (1966) ;̂ ® H'/i/t-corr)b v. Chavis. However, because the underlying principle is aconcern for fairness in poiiticai representation, logic dictates thatgerrymandering claims against single-member districts likewiseshould be justiciable. In Caffney v. Cummfngs, although thestate's plan was upheld. Justice White did say that the Court"must . . . respond" to the claim that the plan was "invidiouslydiscriminatory because a 'political fairness principle was followed."^^

846

The question of whether invidious discrimination must beshown by proof of intent to construct a politically slanted district-ing scheme, or may be shown by evidence of a slanted electoraloutcome in actual result, seems to have been resolved in favor ofthe former requirement. When the earlier cases mentioned "effect,"they always conjoined it with discriminatory "purpose." More re-cently in a non reapportionment case, Washington v. Davis (1976),20the Court expressly rejected the idea that allegations of invidiousdiscrimination in violation of the equal protection of the laws princi-ple of the Fourteenth and Fifth Amendments could be sustained bybypassing proof of discriminatory purp>ose and showing only a dis-proportionate effect of the challenged practice. In that case plain-tiffs could not show invidious racial intent in constructing thefederal civil service examination, but did show that blacks failedthe test in higher proportion than did whites.

For the very reason that "intent" proof is difficult andcourts may tend now to be satisfied with any "quite equal" re-districting plan, no matter how badly gerrymandered in actualresult, it becomes all the more important that equality of politicalopportunity (which also can be called fairness or neutrality) beconsidered in the process of constructing the plan. It may be thefirst and last chance. Of course Justice White did offer a provoca-tive line in Caffney on the intent issue when he said:

"[A] pofi'ti'caliy mi'ndiess approach may produce, whetherintended or not, the most grossly gerrymandered re-suits; and, in any event, it is most uniil<eiy that thepoli'ti'caf impact of such a pian wouid remain undiscoveredby the time it was proposed or adopted, in which eventthe results wouid be both known, and, if not changed^intended. "^^ (Emphasis added.)

We may well hear more of this warning in the litigation which willfollow the 1980 census-induced redistrictings. But it would beunwise to place sole reliance on it as a safeguard for representa-tive fairness.

DISTRICTING STANDARDS.

Contiguity. The contiguity requirement — that no part of onedistrict be completely separated from any other part of the samedistrict — has been universally accepted and poses no enforcementproblem or serious challenge to districting flexibility in pursuit ofother fair representation values.

Compactness. The requirement of compactness specifies that theboundaries of each district shall be as short as practicable. Al-though there is no federal constitutional requirement of compact-ness, such a requirement may present a certain restraint ongerrymandering, and may seem innocuous on its face. Rigid ad-herence to a compactness, however phrased, should be avoided.A district pattern of symmetrical squares, although conceivable,could well operate to submerge a significant element of the elector-ate. As a practical matter, absolute compactness (districts form-ing perfect circles which are even shorter lined than squares) is an

847

impossibility. Furthermore, a benign gerrymander, in the senseof some asymmetrical districts, may well be required in order toassure representation of submerged elements within a larger area.Shape requirements focus on form rather than the substance ofeffective poiiticai representation.

Following Political Boundaries. The requirement of honoring po-iitical subdivision boundaries insofar as possible under the popu-lation equality requirements responds to a traditional and even in-stinctive sense of "community" as a significant basis for represen-tation. It was mentioned by Chief Justice Warren in Reynolds v.Sims, The extent to which political subdivision boundaries may behonored is, of course, an inverse corollary of the degree of popu-lation stringency required. The Supreme Court's populationequality stringency in Kirkpatrick v. Preisler and White v. Weiserhad the effect of badly trampling any political subdivision policyfor congressional districting.

Population Deviation. A 2% maximum deviation rule would leavesome room for following political subdivision lines, without anydemonstrable cost to any principle of fair and effective represen-tation of which I am aware. In Weiser in 1973 the court voidedan official state plan with a maximum deviation of 2.43%, under theforce of the Kirkpatrick rule. The alternative plaintiffs' planadopted by the Court cut 18 more county lines than did the re-jected state plan. I have already noted above that three membersof the Weiser Court said they were disenchanted with this ex-tremely stringent rule, and two additional members had at leastopposed the rule at its inception in 1969.

Incumbent Protection. Various groups (most notably CommonCause, 1977), have advocated a flat prohibition on the considera-tion of addresses of incumbents in the drawing of district lines.In practice this prohibition may be no more enforceable than Pro-hibition itself, but I will lay that aside. From the standpoint ofconstitutional law, the Supreme Court said in Weiser, as it hadearlier, that: "The fact that district boundaries may have beendrawn in a way that minimized the number of contests betweenpresent incumbents does not in and of itself establish invidious-ness. "23 On the question whether the prohibition is sound policy,persons will differ. There are virtues in having some continuityin office for the sake of experience, stability, and relations withconstituents. There also are virtues in turnover, which isprobably far better achieved, if we are at all serious about i t , bylimiting legislators to a specified number of terms.

Dilution of Minority Voting Strength. Rulings like the one madeby the Supreme Court in United Jewish Organizations v. Careysuggest the likelihood that in many states. North as well as South,much of the state legislative reapportionment and congressionaldistricting after the 1980 census will be controlled by the AttorneyGeneral of the United States under the Voting Rights Act. Underthe United Jewish Organizations precedent the control would takethe form of creation of safe districts for a specified number ofblack and language minority candidates, with consequent problems

848

of adverse impacts on other groups. In that case the HasidicJewish community in Brooklyn was split in an attempt to createsafe black districts.

CONCLUSION. In closing I must state that to me districtingmethod is more important than districting standards. In general Ifavor a bipartisan commission with tie breaker. ^̂ ^ Such a provisionhas much merit for a great majority of states. There may be somedifficulty, and I suppose opposition, in heavily one-party states.Even there, however, there would be virtue in getting the district-ing process out in the open through the bipartisan commission de-vice so that it can be observed, and alternative plans adequatelytested. Caffney v. Cummings indicates that there is no federalconstitutional barrier to use of bipartisan commissions, or use bysuch commissions of data on political behavior where done for thepurpose of avoiding piolitically unfair districts.

It is important that the operation of a bipartisan commissionnot be unduly impeded by too detailed a specification of "stan-dards," or attempted limitation on the kind of data that can beconsidered. A straitjacketed commission may be worse than nocommission at all. A policy of blindly choosing one of the greatmany equally equal plans a cartographer or computer operator mayproduce, coupled with the unlikelihood of effective judicial policingof unfair results, could enshrine gerrymandering — perhaps un-planned but none the better for that — in an impregnable position.

Let me illustrate the point by reference one more time to theunfolding of the Connecticut case, Caffney v. Cummings, back in1972. While I was just beginning to take over the case and get aSupreme Court stay (which was granted) of the federal districtcourt ruling adverse to the state's plan, the district court con-currently was proceeding to reapportion the state itself with theaid of a master. The appointed master was Professor Robert H.Bork of Yale University Law School. He worked with graduatestudents and a computer under severe time pressure, and underthe instructions of the district court essentially utilized censusdata alone. His resultant plan, though it certainly was not hisintent, was so favorable to the Democratic party that, I am told,he was specially congratulated by the state Democratic chairman,John Bailey. Absent the continued litigation in the Supreme Court,this plan might well have gone into effect. ^^ There is no substi-tute for using all the knowledge available, and testing all proposedplans with all the knowledge available.

NOTES

1. Baker v. Carr (1962) 369 U.S. 186.

2. Baker v. Carr (1962) 369 U.S . 186 at 269.

3. United Jewish Organizations v . Carey (1977) 430 U.S.144.

4. Whitcomb v. Chavis (1971) 403 U.S. 124.

5. Gaffney v. Cummings (1973) 412 U.S. 735.849

6. Limitations of this sort on the use of political data seemto be envisioneci in Common Cause (1977).

7. Colegrove v. Green (1946) 328 U.S. 549 at 553-54.

8. Wesberry v. Sanders (1964) 376 U.S . 1.

9. Reynolds v. Sims (1964) 377 U.S. 533.

10. Reynolds v. Sims (1964) 377 U.S. 533 at 577.

11. Reynolds v. Sims (1964) 377 U.S. 533 at 578-79.

12. Reynolds v . Sims (1964) 377 U.S. 533 at 565-66.

13. Kirkpatrick v . Preisler (1969) 394 U.S. 526 at 530-31.

14. WeUs V. Rockefeller (1969) 394 U.S. 542.

15. White V. Weiser (1973) 412 U.S. 783.

16. Mahan v . HoweU (1973) 410 U.S. 315.

17. Fortson v. Dorsey (1965) 379 U.S. 433.

18. Burns v. Richardson (1966) 384 U.S. 73.

19. Gaffney v. Cummings (1973) 412 U.S. 735 at 751-51.

20. Washington v. Davis (1976) 426 U.S. 229.

21. Gaffney v. Cummings (1973) 412 U.S. 735 at 753.

22. Gaffney v. Cummings (1973) 412 U.S. 735 at 752, N. 18.

23. White v . Weiser (1973) 412 U.S. 783 at 797.

24. Editors' Note: As additional material in his testimony,not included in this paper, makes clear, Dixon was not advocatinga 2% maximum deviation stan(iard, but rather arguing that such aguideline was superior to a rigid and impractical insistence on"perfect" equality of district population.

25. For (details on bipartisanship and political fairness in the1972 Connecticut reapportionment plan, see Gaffney v. Cummings(1973) 412 U.S. 735 Brief for Appellant, 45 et seq. and ReplyBrief for Appellant (with tables testing state plan, plaintiff'sthree plans, master's plan).

REFERENCES

CCX, A (1966) "Constitutional adjudication anci the promotion ofhuman r ights ," Harvard Law Review 80:91.

COMMON CAUSE (1977) "Toward a system of fair and effectiverepresentation." Report (November).

DIXON, R. (1969) "The Warren Court crusade for the Holy Grailof 'one man, one vote," The Supreme Court Review. Chicago:University of Chicago Press .

NIEMI, R. and J. DEEGAN (1978) "A theory of political district-ing," American Political Science Review 72 (December): 1304-1323.

850