with whom Mr. Justice Douglas and Mr. Justice Marshall concur, dissenting.
The Court's opinion in these cases, and that in American Party of Texas v. White, post, p. 767, hold — correctty *756in my view — that the test of the validity of state legislation regulating candidate access to the ballot is whether we can conclude that the legislation, strictly scrutinized, is necessary to further compelling state interests. See ante, at 736; American Party of Texas v. White, post, at 780-781; for, as we recognized in Williams v. Rhodes, 393 U. S. 23, 30 (1968), such state laws “place burdens on two different, although overlapping, kinds of rights- — the right of individuals to associate for the advancement of political beliefs, and the right of qualified voters, regardless of their political persuasion, to cast their votes effectively.” The right to vote derives from the right of association that is at the core of the First Amendment, protected from state infringement by the Fourteenth Amendment. NAACP v. Button, 371 U. S. 415, 430 (1963); Bates v. Little Rock, 361 U. S. 516, 522-523 (1960); NAACP v. Alabama, 357 U. S. 449, 460-461 (1958). Indeed, the right to vote is “a fundamental political right, because preservative of all rights,” Yick Wo v. Hopkins, 118 U. S. 356, 370 (1886), and “[o]ther rights, even the most basic, are illusory if the right to vote is undermined,” Wesberry v. Sanders, 376 U. S. 1, 17 (1964). See also Reynolds v. Sims, 377 U. S. 533, 555 (1964). Thus, when legislation burdens such a fundamental constitutional right, it is not enough that the legislative means rationally promote legitimate governmental ends. Rather,
“governmental action may withstand constitutional scrutiny only upon a clear showing that the burden imposed is necessary to protect a compelling and substantial governmental interest. Shapiro v. Thompson, 394 U. S. [618, 634 (1969)]; United States v. Jackson, 390 U. S. 570, 582-583 (1968); Sherbert v. Verner, 374 U. S. 398, 406-409 (1963). And once it be determined that a burden has been *757placed upon a constitutional right, the onus of demonstrating that no less intrusive means will adequately protect compelling state interests is upon the party seeking to justify the burden. See Speiser v. Randall, 357 U. S. 513, 525-526 (1958).” Oregon v. Mitchell, 400 U. S. 112, 238 (1970) (separate opinion of Brennan, White, and Marshall, JJ.).
See also Dunn v. Blumstein, 405 U. S. 330, 336-337 (1972); Kramer v. Union Free School District, 395 U. S. 621, 627 (1969); Williams v. Rhodes, 393 U. S., at 31.
I have joined the Court’s opinion in American Party of Texas v. White, supra,1 because I agree that, although the conditions for access to the general election ballot imposed by Texas law burden constitutionally protected rights, nevertheless those laws “are constitutionally valid measures, reasonably taken in pursuit of vital state objectives that cannot be served equally well in significantly less burdensome ways.” Post, at 781. I dissent, however, from the Court’s holding in these cases that, although the California party disaffiliation rule, Cal. Elections Code § 6830 (d) (Supp. 1974), also burdens constitutionally protected rights, California’s compelling state interests “cannot be served equally well in significantly less burdensome ways.”
I
The California statute absolutely denies ballot position to independent candidates who, at any time within 12 months prior to the immediately preceding primary election, were registered as affiliated with a qualified political party. Intertwined with Cal. Elections Code §§2500-2501 (1961), which require primary elec*758tions to be held five months before the general election, § 6830 (d) (Supp. 1974) plainly places a significant burden upon independent candidacy — and therefore effectively burdens as well the rights of potential supporters and voters to associate for political purposes and to vote, see Williams v. Rhodes, supra, at 30; Bullock v. Carter, 405 U. S. 134, 143 (1972) — because potential independent candidates, currently affiliated with a recognized party, are required to take affirmative action toward candidacy fully 17 months before the general election. Thus, such candidates must make that decision at a time when, as a matter of the realities of our political system, they cannot know either who will be the nominees of the major parties, or what the significant election issues may be. That is an impossible burden to shoulder. We recognized in Williams v. Rhodes, supra, at 33, that “the principal policies of the major parties change to some extent from year to year, and . . . the identity of the likely major party nominees may not be known until shortly before the election . . . .” Today, not even the casual observer of American politics can fail to realize that often a wholly unanticipated event will in only a matter of months dramatically alter political fortunes and influence the voters’ assessment of vital issues. By requiring potential independent candidates to anticipate, and crystallize their political responses to, these changes and events 17 months prior to the general election, § 6830 (d) (Supp. 1974) clearly is out of step with “the potential fluidity of American political life,” Jenness v. Fortson, 403 U. S. 431, 439 (1971), operating as it does to discourage independent candidacies and freeze the political status quo.
The cases of appellants Storer and Frommhagen pointedly illustrate how burdensome California’s party disaffiliation rule can be. Both Storer and Frommhagen sought to run in their respective districts as inde*759pendent candidates for Congress. The term of office for the United States House of Representatives, of course, is two years. Thus, § 6830 (d) (Supp. 1974) required Storer and Frommhagen to disaffiliate from their parties within seven months after the preceding congressional election. Few incumbent Congressmen, however, declare their intention to seek re-election seven months after election and only four months into their terms. Yet, despite the unavailability of this patently critical piece of information, Storer and Frommhagen were forced by § 6830 (d) (Supp. 1974) to evaluate their political opportunities and opt in or out of their parties 17 months before the next congressional election.
The Court acknowledges the burdens imposed by § 6830 (d) (Supp. 1974) upon fundamental personal liberties, see ante, at 734, but agrees with the State’s assertion that the burdens are justified by the State’s compelling interest in the stability of its political system, ante, at 736. Without § 6830 (d) (Supp. 1974), the argument runs, the party’s primary system, an integral part of the election process, is capable of subversion by a candidate who first opts to participate in that method of ballot access, and later abandons the party and its candidate-selection process, taking with him his party supporters. Thus, in sustaining the validity of § 6830 (d) (Supp. 1974), the Court finds compelling the State’s interests in preventing splintered parties and unrestricted factionalism and protecting the direct-primary system, ante, at 736.2
*760But the identification of these compelling state interests, which I accept, does not end the inquiry. There remains the necessity of determining whether these vital state objectives “cannot be served equally well in significantly less burdensome ways.” Compelling state interests may not be pursued by
“means that unnecessarily burden or restrict constitutionally protected activity. Statutes affecting constitutional rights must be drawn with ‘precision,’ NAACP v. Button, 371 U. S. 415, 438 (1963); United States v. Robel, 389 U. S. 258, 265 (1967), and must be ‘tailored’ to serve their legitimate objectives. Shapiro v. Thompson [394 U. S. 618, 631 (1969)]. And if there are other, reasonable ways to achieve those goals with a lesser burden on constitutionally protected activity, a State may not choose the way of greater interference. If it acts at all, it must choose ‘less drastic means.’ Shelton v. Tucker, 364 U. S. 479, 488 (1960).” Dunn v. Blumstein, 405 U. S., at 343.
While it is true that the Court purports to examine into “less drastic means,” its analysis is wholly inadequate. The discussion is limited to these passing remarks, ante, at 736:
“Nor do we have reason for concluding that the device California chose, § 6830 (d) (Supp. 1974), was not an essential part of its overall mechanism to achieve its acceptable goals. As we indicated in Rosario, the Constitution does not require the State to choose ineffectual means to achieve its aims. To conclude otherwise might sacrifice the political stability of the system of the State, with profound con*761sequences for the entire citizenry, merely in the interest of particular candidates and their supporters having instantaneous access to the ballot.”
Naturally, the Constitution does not require the State to choose ineffective means to achieve its aims. The State must demonstrate, however, that the means it has chosen are “necessary.” Shapiro v. Thompson, 394 U. S. 618, 634 (1969). See also American Party of Texas v. White, post, at 780-781.
I have searched in vain for even the slightest evidence in the records of these cases of any effort on the part of the State to demonstrate the absence of reasonably less burdensome means of achieving its objectives. This crucial failure cannot be remedied by the Court’s conjecture that other means “might sacrifice the political stability of the system of the State” (emphasis added). When state legislation burdens fundamental constitutional rights, as conceded here, we are not at liberty to speculate that the State might be able to demonstrate the absence of less burdensome means; the burden of affirmatively demonstrating this is upon the State. Dunn v. Blumstein, supra, at 343; Shapiro v. Thompson, supra, at 634; Sherbert v. Verner, 374 U. S. 398, 406-409 (1963).
Moreover, less drastic means — which would not require the State to give appellants “instantaneous access to the ballot” — seem plainly available to achieve California’s objectives. First, requiring party disaffiliation 12 months before the primary elections is unreasonable on its face. There is no evidence that splintering and factionalism of political parties will result unless disaffiliation is effected that far in advance of the primaries. To the contrary, whatever threat may exist to party stability is more likely to surface only shortly before the primary, when the identities of the potential field of candidates and issues *762become known. See Williams v. Rhodes, 393 U. S., at 33. Thus, the State’s interests would be adequately served and the rights of the appellants less burdened if the date when disaffiliation must be effected were set significantly closer to the primaries. Second, the requirement of party disaffiliation could be limited to those independent candidates who actually run in a party primary. Section 6830 (d) (Supp. 1974) sweeps far too broadly in its application to potential independent candidates who, though registered as affiliated with a recognized party, do not run for the party’s nomination. Such an independent candidate plainly poses no threat of utilizing the party machinery to run in the primary, and then declaring independent candidacy, thereby splitting the party.
II
I also dissent from the Court’s remand, in the case of appellants Hall and Tyner, of the question concerning the constitutionality of the petition requirements imposed upon independent candidates. Under the relevant statutes, Hall and Tyner, candidates for President and Vice President, were required to file signatures equal to 5% of the total vote cast in California’s preceding general election. § 6831. However, the pool from which signatures could be drawn excluded all persons who had voted in the primary elections, including voters who had cast nonpartisan ballots. § 6830 (c) (Supp. 1974). Furthermore, circulation of the petitions was not permitted until two months after the primaries, and the necessary signatures were required to be obtained during a 24-day period. § 6833 (Supp. 1974). The Court avoids resolving the constitutionality of these election laws by remanding to the District Court for further proceedings. On remand, the District Court is directed to determine (1) the total vote cast in the last general election as a predi*763cate to computation of the 5% of signatures required by the statutory provision, and (2) the size of the pool to which appellants were required to limit their efforts in obtaining signatures. The Court reasons that these findings are necessary to a determination “whether the available pool is so diminished in size by the disqualification of those who voted in the primary that the 325,000-signa-ture requirement, to be satisfied in 24 days, is too great a burden -on the independent candidates for the offices of President and Vice President.” Ante, at 740.
If such a remand were directed in the cases of Storer and Frommhagen I could agree, for in those cases there is a complete absence of data necessary to facilitate determination of the actual percentage of available voters that appellants Storer and Frommhagen were required to secure. A remand in the case of Hall and Tyner, however, is unnecessary because the data upon which relevant findings must be based are already available to us. The data are cited by the Court, ante, at 742 n. 12 and at 744 n. 14. Evaluated in light of our decision in Jenness v. Fortson, supra, the data leave no room for doubt that California’s statutory requirements are unconstitutionally burdensome as applied to Hall and Tyner. Official voting statistics published by the California Secretary of State indicate that 6,633,400 persons voted in the 1970 general election. See Secretary of State, Statement of Vote, General Election, November 7, 1972, p. 6. Appellants were required to secure signatures totaling 5% of that number, i. e., 331,670. The statistics also indicate the size of the total pool from which appellants were permitted to gather signatures. The total number of registered voters on September 14, 1972 — the last day appellants were permitted to file nomination petitions — was 9,953,124. See Secretary of State, Report of Registration, September 1972, p. 8. Of that number, 6,460,220 *764registered voters could not sign petitions because they had voted in the 1972 primary elections. See Secretary of State, Statement of Vote, Consolidated Primary Election, June 6, 1972, pp. 3, 4-23. Thus, the total pool of registered voters available to appellants was reduced to approximately 3,492,904, of which the required 331,670 signatures was 9.5%.3
In my view, a percentage requirement even approaching the range of 9.5% serves no compelling state interest which cannot be served as well by less drastic means. To be sure, in Jenness we acknowledged that:
“There is surely an important state interest in requiring some preliminary showing of a significant modicum of support before printing the name of a political organization’s candidate on the ballot — the interest, if no other, in avoiding confusion, deception, and even frustration of the democratic process at the general election.” 403 U. S., at 442.
We there upheld the constitutionality of Georgia’s election laws requiring potential independent candidates to gather the signatures equal to 5% of the total eligible electorate at the last general election for the office in question. However, candidates were given a full six months to circulate petitions and no restrictions were placed upon the pool of registered voters from which *765signatures could be drawn. In that circumstance, we found that Georgia imposed no unduly burdensome restrictions upon the free circulation of nominating petitions. We noted:
“A voter may sign a petition even though he has signed others, and a voter who has signed the petition of a nonparty candidate is free thereafter to participate in a party primary. The signer of a petition is not required to state that he intends to vote for that candidate at the election. A person who has previously voted in a party primary is fully eligible to sign a petition, and so, on the other hand, is a person who was not even registered at the time of the previous election. No signature on a nominating petition need be notarized.” Id., at 438-439 (footnotes omitted).
Thus, although Georgia’s 5% requirement was higher than that required by most States, the Court found it “balanced by the fact that Georgia . . . imposed no arbitrary restrictions whatever upon the eligibility of any registered voter to sign as many nominating petitions as he wishes.” Id., at 442.
California seeks to justify its election laws by pointing to the same substantial interests we identified in Jenness, of insuring that candidates possess a modicum of support, and that voters are not confused by the length of the ballot. But in sharp contrast to the election laws we upheld in Jenness, California’s statutory scheme greatly restricted the pool of registered voters from which appellants Hall and Tyner were permitted to draw signatures. The 5% requirement, in reality, forced them to secure the signatures of 9.5% of the voters permitted by law to sign nomination petitions. Moreover, unlike Georgia’s six-month period for gathering signa*766tures, the California election laws required appellants to meet that State’s higher percentage requirement in only 24 days. Thus, even conceding the substantiality of its aims, the State has completely failed to demonstrate why means less drastic than its high percentage requirement and short circulation period — such as the statutory scheme enacted in Georgia — will not achieve its interests.
Accordingly, I would reverse the judgment of the District Court dismissing these actions, and remand for further proceedings consistent with this opinion.
Mr. Justice Douglas adheres to the views stated in his opinion dissenting in part in American Party of Texas v. White, post, p. 795.
The Court also opines that § 6830 (d) (Supp. 1974) may be “a substantial barrier to a party fielding an ‘independent’ candidate to capture and bleed off votes in the general election that might well go to another party,” ante, at 735. But the State suggests no reliance upon this alleged interest and we are therefore not at liberty to turn our decision upon our conjecture that this might have been a state objective. In any event, the prospect of such a misuse seems more fanciful than real and, as we said in Williams v. Rhodes, 393 U. S. *76023, 33 (1968), “[n]o such remote danger can justify [an] immediate and crippling impact on . . . basic constitutional rights
The Court’s computations, ante, at 744 n. 14, suggest that Hall and Tyner need only have collected signatures from 8.1% of the available voter pool. The Court’s calculation assumes that the voter pool available to Hall and Tyner included approximately 579,000 persons who may have only voted in nonpartisan primaries. Section 6830 (c) (Supp. 1974) makes no such exception; the pool available for signatures is expressly limited to those voters who “did not vote at the immediately preceding primary election . . . .” I agree with the Court, however, that exclusion of persons voting at nonpartisan primaries is not supported by a compelling state interest.