concurring in part and dissenting in part:
I dissent from so much of the judgment and the majority opinion as affirms dismissal of the salary and 4-H Club claims. As to both of those I would find reversible error and remand for further proceedings. I concur in the balance of the judgment.
I
With regard to the salary discrimination claim, I would hold that the district court erred in its refusal to certify the class ultimately sought by the private plaintiffs; in its decision on the merits that no discriminatory “pattern or practice” had been proved; and in its rejection of the individual claims following its “no pattern or practice” conclusion.
A.
The individual plaintiffs’ challenge to the district court’s refusal to certify a class consisting of “all black employees of [the Extension Service] on and after November 18, 1971,” the date of commencement of the action, is well taken.
There was no possible basis for declining ultimately to certify such a class.
It clearly met the numerosity requirement. The number of black professional employees has been more than one hundred in each of the years involved. The proposed class would be even larger because of turnover during these years. These figures were undisputed.
The other basic criteria for class certification — commonality of issues, typicality of claims, adequacy of representation — were equally well established. The district court’s refusal, early and late, to certify the class action was, however, obviously based upon a perception that these criteria were not met. Its stated reasons were two. Neither bears scrutiny.
The first was the court’s apparent view, confirmed finally by its “no pattern or practice” conclusion on the merits, that the locus of decisions on salary matters was so far decentralized to the county level that a class action alleging a pattern or practice multi-county in its reach was not warranted. Citing to this court’s decision in Hill v. Western Electric Co., 596 F.2d 99, 102 (4th Cir.1979) (Hill I), the court concluded that “in employment discrimination cases ... all the plaintiff representatives and all the class members must be from the same work facility ...” and that here “the representative plaintiffs ... come from roughly thirty of the one hundred counties,” so that “[i]n no sense can they be deemed to work within the same facility____” App. 24.
This misconceived the holding of Hill I. That case laid down no per se rule preventing class certification in any multi-facility case (even assuming that the employment relationship here is sufficiently like that in the multi-facility industrial setting of Hill to warrant comparable general treatment). Hill simply applied to the multi-facility setting of that case the general rule that claims of representative plaintiffs must be typical of those of the class, that the representatives must also be situated adequately to represent the interests of class members, and that these criteria, along with the commonality of issues criterion, were not met where multiple facilities were essentially autonomous in respect of the employment decisions directly in issue. This was made plain in our later decision in Stastny v. Southern Bell Telephone and Telegraph Co., 628 F.2d 267 (4th Cir.1980). There, though certification was again found improper in a multi-facility setting, it was emphasized that the reason was not the setting itself, but the fact of facility autonomy. Id. at 277, 279 & n. 19.
Here, unlike the situations in both Hill I and Stastny, there was abundant proof, both preliminarily and increasingly during the course of trial on the merits, that despite local county contributions to salaries, hence some county-level participation in the salary decisions in issue, the dominant control over total salary levels lay ultimately in the state-level Extension Service. Thus, although it was undisputed that the level of *689county contributions to the salaries in question varied over the state, it was also clear from the evidence that ultimate salary level determinations, even of those putative class members working at the county level, were, as a matter of standard practice, subject to approval by Extension Service officials. This is revealed most dramatically in the recognition by Extension Service. officials in 1971 that the race-based salary gaps dating back to pre-1965 overtly discriminatory practices must in prudence and by law now somehow be closed. Those officials plainly recognized and acted upon this necessity as a matter of Extension Service responsibility.
From this it was clear from the outset of this litigation, and only became clearer as it proceeded, cf. Stastny, 628 F.2d at 276, that the dominant issue of fact and law respecting the salaries of putative class members was not whether there was significant state-level control over those salary levels; that was beyond dispute. The only possibly disputable issue was whether that control was exercised in racially discriminatory ways in accordance with a pattern or practice similarly affecting all members of the putative class. Indeed, in the final analysis, the district court, in rejecting the “pattern or practice” claims advanced by the individual plaintiffs, recognized the claim’s unmistakably appropriate class-action basis by referring to it as a “class-claim.”
Which leads to the second, and probably more critical, basis for the district court’s refusal to certify a salary-claim class. This was the court’s stated perception that once the Government had intervened with its own “pattern or practice” claim respecting salaries, the private party class action question was essentially mooted. As the district court saw it, the government’s claim, if meritorious, would be as efficacious to the putative class members as would have been a private class action in their behalf.
General Telephone Co. v. EEOC, 446 U.S. 318, 100 S.Ct. 1698, 64 L.Ed.2d 319 (1980), directly refutes this. General Telephone held that a government pattern or practice suit need not be structured as a Rule 23 class action precisely because to do so would preclude, as class members, individuals who might prefer, and whom Congress intended could, proceed in individual or private class actions, seeking parallel relief. Id. at 333, 100 S.Ct. at 1707.
It was therefore prejudicial error to decline to certify the salary-claim class sought by plaintiffs.
B.
Whether analyzed from the perspective of the Government’s or the private plaintiff’s “pattern or practice” claims,1 I would then also find reversible error in the district court’s rejection of that claim on the merits.
(D
It is helpful at the outset to recall the now familiar features of the two-stage proceeding that has developed under Title YII for the prosecution of “pattern or practice” claims of disparate treatment, whether by private class action or government enforcement action.
In the first stage, the plaintiff need only establish by a preponderance of the evidence that a pattern or practice of disparate treatment has existed as a matter of an employer’s regular and standard operating procedure. Hazelwood School District v. United States, 433 U.S. 299, 307, 97 S.Ct. 2736, 2741, 53 L.Ed.2d 768 (1977); Teamsters v. United States, 431 U.S. 324, 336, 97 S.Ct. 1843, 1855, 52 L.Ed.2d 396 (1977); EEOC v. Ford Motor Co., 645 F.2d 183, 197-98 (4th Cir.1981), rev’d on other grounds, 458 U.S. 219, 102 S.Ct. 3057, 73 L.Ed.2d 721 (1982). At this stage, the plaintiff need not attempt proof that any particular employees are specific victims of the pattern or practice in order to establish *690its existence. Teamsters, 431 U.S. at 360, 97 S.Ct. at 1867. Rather, as is customary, the plaintiff may rely primarily upon statistical evidence, bolstered by such other evidence as he may care to offer. See, e.g., EEOC v. American National Bank, 652 F.2d 1176, 1188 (4th Cir.1981).
Statistical evidence, either alone or in conjunction with other evidence, may carry the plaintiffs burden at this stage, though it is of course subject to rebuttal by demonstration of methodological flaws or inaccuracies. See id. If this burden is carried, the resulting finding of a discriminatory pattern or practice gives rise to an inference that all employees subject to the policy were its victims and are entitled to individually appropriate remedies. Teamsters, 431 U.S. at 362, 97 S.Ct. at 1868; Sledge v. J.P. Stevens & Co., 585 F.2d 625, 637 (4th Cir.1978). This leads to the second stage, in which individual claimants may attempt proof of their entitlement to individual remedies. Teamsters, 431 U.S. at 361-62, 97 S.Ct. at 1867-68. In this second stage individual claimants are entitled to a presumption that they are victims of the discriminatory pattern or practice established in Stage I, and the burden is thereupon on the employer to prove the contrary. Teamsters, 431 U.S. at 359 n. 45, 97 S.Ct. at 1867 n. 45; Sledge v. J.P. Stevens & Co., 585 F.2d 625, 637 n. 25 (4th Cir.1978).
(2)
In this case, the district court rejected the Stage I pattern or practice claim and, in consequence, never reached Stage II proceedings. In doing so, its finding that no cognizable pattern or practice of intentional racial discrimination in salaries had been established was clearly erroneous. This error is traceable to certain misapprehensions of controlling legal principles and to a failure by the court to accord any weight to certain evidence strongly probative of the claim. See Miller v. Mercy Hospital, 720 F.2d 356, 361 (4th Cir.1983).
Consideration must start with the undisputed fact that prior to the 1965 merger of white and Negro branches of the Extension Service, the salaries of black professionals were intentionally and quite openly simply set lower than those of white colleagues in the same employment positions. The differential was substantial at all levels and was maintained as a matter of general policy that reflected the cold fact — as testified by an Extension Service official at trial — that at that time blacks could simply be hired and retained at lower salaries than could whites in comparable positions.
Following the 1965 merger responsive to the Civil Rights Act of 1964, entry-level base salaries for whites and blacks in comparable professional positions were equalized. But substantial average differentials in all positions continued to exist, and the evidence clearly showed that in time these were reflected in the salaries of both pre1965 and post-1965 hires. As to the former, because the pre-1965 gaps in base salaries had not been closed; as to the latter, because a significant gap developed once post-1965 hires had been employed long enough to move up from entry level positions. That these dual strands of substantial salary differentials continued well past 1965 and, more critically, well past 1968 (the earliest limitation date applicable to the salary claim2) was established by undisputed evidence.
Perhaps most dramatic was an internal memorandum prepared in 1971 by Dr. Blal-ock, then assistant director of the Extension Service budget, and an employee of the Service in pre-merger days. Presumably as a result of inquiries or self-prompted concerns about the implications of federal law for the continued maintenance of salary differentials, the memorandum first flatly conceded that “salaries for women and non-white men are lower. Our figures verify.” This was attributed to several *691factors: “the competitive market” (recognized as “not acceptable as a reason though”); “tradition”; and “less county support for non-white positions.”
. Based on this, Blalock had urged in 1971 that action should be taken “as quickly as possible] and preferably before our plan went into Washington.” He warned:
Obviously one of the areas where we’ll be checked is on salary. Easy to measure and at least see if there appears to be any [differential].
Blalock had then in fact gone so far as to calculate the salary adjustments needed to close the black/white salary differentials,3 concluding that increases averaging in the range of $800 to $1100 for black professionals would be required.
There is no dispute that Blalock’s general assessment of average salary differentials still existing in 1971 was accurate. But the evidence further disclosed, again without dispute, that though some general adjustments were then put into effect, they did not approach those calculated by Blal-ock as required to close the gaps. The adjustments were not in fact significant so far as redressing existing race-based differentials. For example, plaintiffs’ evidence revealed, again without dispute, that the average gap between white and black agricultural agents at all ranks was then reduced by only $85; that the gap between black and white home economics agents actually increased by $13; and that in three of the six categories of professional employees of the Extension Service the average differentials increased in 1971 over those obtaining in 1970. Approximately a year and a half after the 1971 adjustments, an internal analysis by the Extension Service of the salaries being paid its full agents revealed that those of blacks still averaged $455 less than those paid whites with comparable educational levels and tenure.
From 1971 forward, average differentials between the salaries paid black and white professionals continued to decrease. But the evidence is clear that those average differentials were not substantially removed for home economics agents until 1974 and for agricultural agents until 1976.4
That these differentials persisted well into the applicable limitations periods was recognized by the district court, as the evidence compelled that it must be. The court observed of the 1971 “adjustments” —the first undertaken since 1965 — that they only began to eliminate the differentials that had originated in pre-1965 overt discriminatory policies. And the district court recognized that the pattern of disparities continued well past that time.
There remained of course the question of motive — whether the clearly established pattern of salary differentials disfavoring blacks was, during the period of its post-1968 existence, “racially premised.” See Teamsters, 431 U.S. at 335, 97 S.Ct. at 1854. While that motive might possibly have been inferred “from the mere fact of [the] differences in treatment,” id. n.15, both sides presented additional evidence of motive in the form of regression analyses of statistical data.
For the government as plaintiff, Dr. Charles Mann presented regression analy-ses he had done for the years 1974, 1975 and 1981. Certain of these used four independent variables to assess effects upon salaries — race, education, tenure, and job title. His analyses demonstrated statistically significant racial effects for 1974 and 1975, and statistically insignificant racial effects for 1981. Specifically, it was demonstrated that in 1974 the average black professional earned $331 less than a white professional with the same education, ten*692ure and job title; in 1975, $335 less; and in 1981, still $248 less.
A regression analysis commissioned by the Extension Service was entirely consistent with that of Dr. Mann for the same time period. Using the same independent variables, it demonstrated a racial effect of $364-$381 disfavoring black professionals as of October 1974.
Dr. Giesbrecht, an expert witness for the defendants, presented multiple regression analyses for 1975 and 1981. His results were also consistent with those of Dr. Mann. Using race, education, tenure and job level as independent variables, they demonstrated a statistically significant racial effect of $384 for the year 1975; a statistically insignificant racial effect of $310 for 1981. Giesbrecht also presented regression analyses for the same years that added as an independent variable “quartile rankings” that reflected job performance evaluation. For 1975, using this variable in addition to the others increased the race effect to $475; for 1981, it decreased the effect to a still statistically insignificant $142.
Notwithstanding the consistency of these analyses and the testimony of the experts on both sides that they demonstrated statistically significant salary differentials attributable to race that continued at least to some time after 1975, the district court declined to accord them any probative force.5 The court reasoned, adopting a specific argument by defendants, that the analyses were fatally flawed by their failure to include nine other independent variables posited by defendants as potentially relevant.6 At the same time, the court apparently rejected as irrelevant to the issue of motive any evidence other than the regression analysis offered by Dr. Mann. This appears from the court’s conclusion that to the extent plaintiffs had established a prima facie case “it has only been by virtue of ... statistical evidence based on the multiple regressions of Dr. Mann.” Finally, the court concluded that, in any event, the plaintiffs’ statistical evidence was undermined by defendants’ countering proof of the absence of significant salary differentials.
In all three of these critical determinations, the court was so clearly in error that its ultimate finding of no pattern or practice should not be allowed to stand.
As to the rejection of Mann’s regression analyses (and along with it that of defendants’ expert) for failure to include a number of other independent variables merely hypothesized by defendants, I completely agree with the government and the private plaintiffs that to apply such a rule generally would effectively destroy the ability to establish any Title VII pattern or practice claim by this means of proof. There is no authority of which I am aware that applies such a rule. As those parties rightly point out, it will always be possible for Title VII defendants to hypothesize yet another vari*693able that might theoretically reduce a race-effect coefficient demonstrated by any multiple regression analysis that could be conceived. Merely hypothesizing other possible variables, no matter how rationally, should not suffice where, as here, a statistically significant race coefficient has been demonstrated by a regression analysis employing the most obvious alternative variables of tenure, education, and job position. Leaving out for the moment the bolstering effect of plaintiffs’ non-statistical evidence of a race-based pattern or practice, the effect of the wholly consistent regression analyses here was unmistakably to make out a prima facie case requiring rebuttal by defendants. See Teamsters, 431 U.S. at 360, 97 S.Ct. at 1867; Chisholm v. United States Postal Service, 665 F.2d 482, 496 (4th Cir.1981). This might have been done, for example, by evidence that the inclusion of other relevant variables would in fact reduce the race-effect coefficient to a statistically insignificant level. But that simply had not occurred here. For all this record discloses, no one of the variables merely hypothesized by the defendants and relied upon by the court would have had any effect on the race-effect coefficient revealed in the regression analyses of both sides’ experts.7
The court also erred in rejecting all evidence other than Mann’s regression analysis as probative of a discriminatory pattern or practice. This dissenting opinion is no place to recapitulate all the other evidence that bolstered the inference of intentional discrimination past 1968. But such evidence abounds in the record. Perhaps most practically compelling is the concession by Blalock in 1971 that the pattern of overt race-based discrimination that originated in pre-merger days was then still in effect, in part at least because of “tradition.” Coupled with this telling concession against interest was that responsible official’s flat warning at the time that remedial action might in prudence be required to avoid difficulty with federal reviewing authorities. It is hard to imagine more weighty evidence of knowing, intentional continuation of acknowledged discriminatory policies.
Other evidence of a purposeful continuation of the practice well past 1971 was presumably disregarded by the district court in view of its stated reliance alone on the Mann regression analysis. Most critical were numerous specific examples of disparate pay between white and black professionals holding the same job positions, having comparable levels of tenure and ed*694ucation, and working in the same county offices at the same time.8
The continued existence of this clear pattern was in fact never sought to be rebutted on any basis that could conceivably have, been thought to explain the pattern itself, as opposed to occasional specific examples. The defendants’ suggestion that somehow merit (as reflected in the quartile rankings), or longevity, or a variance in salary supplements from county to county adequately and fully explained the continued existence of so widespread and substantial a pattern of disparate pay concededly race-based in its origins indeed bordered on the frivolous. On each of these points, the plaintiffs’ evidence included numerous illustrations of pay differential that could not be explained on any of the non-raciál grounds merely hypothesized by defendants.9
Indeed, anyone looking at this whole record, including some of the district court’s own findings and observations could in fairness take only one view of it. It is this: that of course the general pattern of pre-1965 overt discrimination in salary continued in substantial, if gradually diminishing, degree until at least 1976 and perhaps beyond; that of course, the responsible officials of the Extension Service with responsibility and authority in the matter knew that the pattern continued during that time and that to some degree at least it included a carry-over race-based component from its pre-1965 origins; and that of course, so knowing, those officials allowed it to continue for a time when it lay within their power to remedy it.
If this be, as I am persuaded, the only rational assessment to be made of the evidence in this record, the question arises, and in fairness must be addressed: how then did the district court and the majority here fail so to assess it? In my judgment, the answers are readily found in misapprehensions of controlling legal principle, and not in any essential disagreement with my assessment of the facts clearly proven.
I identify three such basic legal misapprehensions by the district court.
The first has already been alluded to in Part IA: the court’s perception that the salary pattern or practice alleged had not been shown to be class-wide in its reach because of county autonomy in salary matters. For reasons there indicated, this view is simply erroneous as a matter of law — whether applied to the class certification or merits issues.
The second has to do with the relevant time frame within which the existence of a pattern or practice of salary discrimination was to be assessed. That time frame commenced at least as far back as 1972 for the Title VII claims; it commenced as far back as 1968 for the constitutional and Title VI claims. While the district court specifically recognized this, its findings and conclusions unmistakably reveal at least a preoccupation, as the plaintiffs suggest, or perhaps a sole concern with the question whether a discriminatory pattern or practice continued to exist at the time of trial in 1981.
For example, the court rejected, presumably as irrelevant, a government comparison of black and white salaries earned in 1973 on the basis that it involved differences existing “about ten years ago,” though this was well within all the applicable limitation periods. Again, the court in discussing the “seeming salary disparaties” revealed by plaintiffs’ regression analyses over the period 1974-1981, concluded that on the basis of defendants’ evidence respecting the years 1976-1981, the seeming disparities had been adequately explained. This defendants’ evidence obviously had nothing to do with the years 1974-1975 covered by plaintiffs’ regression *695analysis; and it certainly spoke not at all to the salary pattern before 1974, a time well within the limitation period. The impression is strongly conveyed that the court simply did not consider that it might be compelled to find a discriminatory pattern or practice of discrimination existing for at least some time after 1972 (or possibly 1968) even though it had been rectified by the time of trial. This of course was an erroneous view of the law.10
The third misapprehension has, possibly, to do with the nature of the discriminatory intent or motive that must be found to establish a pattern or practice claim and, possibly, with legal justifications that might excuse any discriminatory intent found to exist. In a frank and revealing passage, the court opined:
The Extension Service’s problem of bringing black and white salaries into line has been similar to that which faced most business enterprises with a prior history of racial discrimination following the passage of the Civil Rights Act of 1964. Just as it had been found in the area of education that there is no such thing as instant integration, it was soon found in the field of business and industry that there is no such thing as instant equality in employment. Without risking serious disruption of a business by prohibitively costly budgetary alterations and a possible practice of wholesale reverse discrimination it was soon recognized (though not always by the courts) that the adjustments mandated by the law simply could not be made overnight. The dilemma of the Extension Service was further compounded by the fact that its operating funds come from three separate political entities each of which retains a voice in all major employment decisions.
This passage unmistakably reveals, first off, the court’s recognition that a pattern of disparate treatment had been proven; the proferred legal justification or excuse for it that follows would not otherwise be required. Beyond that, it suggests two possible views of the legal significance of the pattern that are equally erroneous in law.
The first is that the intent required to prove a discriminatory pattern or practice could not be found in the mere continuation of a pattern or practice not originated by those presently charged with its existence. The other is that in any event even the intentional continuation of such a pattern or practice would be legally justified by the extreme difficulties of abandoning it.
Whatever the practical wisdom and accuracy of the court’s observations, I agree with plaintiffs’ position that nothing in controlling law supports its legal premises. Only those erroneous premises rather than the court’s actual fact findings could possibly explain the court’s conclusion that no pattern of.illegal salary discrimination existed for periods extending well into covered time periods.
This court’s opinion affirming the district court’s rejection of the salary discrimination claim is, with respect, similarly flawed with basic legal error. Relying principally upon Hazelwood and United Airlines v. Evans, 431 U.S. 553, 97 S.Ct. 1885, 52 L.Ed.2d 571 (1977), the majority holds, in effect, that because the pattern of discriminatory salaries here challenged originated before applicable provisions of the Civil Rights Act made their payment illegal,11 *696any “lingering effects” of that earlier pattern cannot (presumably on an indefinitely maintained basis) be considered in assessing a challenge to post-act continuation of that pattern.12
Hazelwood and Evans indeed made it clear that an employer cannot be found liable, or sanctioned with remedy, for employment decisions made before they were declared illegal or as to which the claimant has lost any right of action by lapse of time. For this reason it is generally true that, as the catch-phrase has it, Title VII imposed “no obligation to catch-up,” i.e., affirmatively to remedy present effects of pre-Act discrimination, whether in composing a work force or otherwise. But those cases cannot be thought to insulate employment decisions that presently are illegal on the basis that at one time comparable decisions were legal when made by the particular employer. It is therefore one thing to say that an employer who upon the effective date of Title VII finds itself with a racially unbalanced work-force need not act affirmatively to redress the balance; and quite another to say that it may also continue to make discriminatory hiring decisions because it was by that means that its present work force was composed. It may not, in short, under the Hazelwood/Evans principle continue practices now violative simply because at one time they were not.
So, in this case, though there might theoretically be some question as to exactly when the challenged post-Act pattern or practice of salary discrimination could be said to have originated,13 there can be no doubt that it was well established in its challenged form within applicable limitation periods. Indeed, with respect to pay and other “condition of employment” claims, as opposed to hiring and other work-force composition claims, it may well be argued that the “no-catch-up” principle simply has no logical application. Certainly, for example, it could not be contended that an employer could continue to maintain segregated dining facilities simply because they originated pre-Act.
B.
As both the Government and the private plaintiffs rightly point out, the district court’s erroneous conclusion that no pattern or practice had been established made also erroneous its denial of the claims of individual plaintiffs. Those plaintiffs, as members of the class- sought to be certified (or as employees properly represented by the Government in its enforcement action) were entitled to have their claims determined with benefit of the “Teamsters presumption,” see Teamsters, 431 U.S. at 359 n. 45, 97 S.Ct. at 1867 n. 45, a benefit denied them by the court’s error. They are entitled, therefore, to have the judgments against them on their individual claims reversed and to participate as claimants in Stage II back-pay proceedings. Sledge v. *697J.P. Stevens & Co., 585 F.2d 625, 637-38 (4th Cir.1978).
C.
Because of the district court’s errors in failing to apply controlling principles of law to the indisputable evidence that defendants knowingly continued a pattern or practice of making and enforcing racially discriminatory salary decisions at least until 1976, I would reverse and remand. The court should be directed to enter judgment declaring that an illegal pattern or practice existed from 1968 until at least 1976, or such later date as the court might now find, and ordering Stage II back-pay proceedings in which individual employees within the class sought to be certified, including the individual plaintiffs whose claims were denied, might prosecute claims with benefit of the Teamsters -presumption. See EEOC v. American National Bank, 652 F.2d at 1201.
Ill
The district court also erred in dismissing on the merits14 plaintiffs’ claim of constitutional and statutory violations in the state’s maintenance of racially segregated 4-H and Extension Homemaker Clubs.
The salient facts, essentially undisputed, are these. The 4-H program is a major Extension Service activity, which receives substantial state and federal funding for whose administration the Extension Service is accountable. One of the program’s critical features is the encouragement, financial support, and servicing of voluntarily organized 4-H and Extension Homemaker Clubs throughout the state.
Before 1965 these clubs were deliberately organized and operated, as was the Extension Service in all its features, on a racially segregated basis. In 1965, following passage of the 1964 Civil Rights Act, the Extension Service began and has since intermittently adopted a series of essentially ineffectual measures facially designed to “integrate” pre-1965 clubs and to avoid racial discrimination in the organization and operation of new clubs, as a part of its general program of mandated compliance with the requirements of Title VI of the Act.
The history of the Extension Service’s adoption, rejection, compromise, interpretation and administration of those anti-discrimination measures is a checkered and confused one. Its ultimate form, however, can fairly be characterized as a freedom of choice approach to the membership composition of these clubs. Affirmative efforts to achieve effective integration of pre-1965 clubs and to avoid segregation in post-1965 clubs have consisted, on any fair assessment of this record, of announcing to Extension Service agents and to the public that the clubs are open to members of all races and forbidding the denial of membership to any applicant on the basis of race.
The actual effects of these desegregation compliance policies have, however, been minimal so far as the racial composition of local clubs is concerned.15 Despite the emergence by 1972 of some new clubs integrated to some extent, the numbers of such clubs in relation to all-black and all-white clubs remained quite low. For example, in that year there were 22 Homemaker Clubs integrated to some degree, but still 1,378 clubs all white and 466 all non-white. Since 1965, the number of all-white 4-H *698Clubs has declined by only 8%; the number of single-race 4-H Clubs in ethnically mixed communities has declined by less than 2% since 1972.
As the plaintiffs rightly point out, the district court did not find, nor do the defendants make any serious claim, that the dual system of single race club membership required by the state before 1965 had been significantly disestablished at the time of trial. The court instead concluded, and the defendants here urge in effect, that this constituted no violation of applicable constitutional or statutory law. I disagree.
The district court essentially concluded that so long as all-white clubs were not proven at the time of trial to be rejecting black applicants on the basis of race, no violation either of Title VI or of the Constitution had been proven. This either misapprehended the nature of the plaintiffs’ claims, or of the applicable law, or both.
The plaintiffs’ claim necessarily has two prongs. The first relates to those present-day single-race clubs that were established before 1965 on a state-imposed segregated basis. As to these, plaintiffs rely on the fourteenth amendment and on a specific Department of Agriculture regulation promulgated under Title VI and having the force of law:
In administering a program regarding which the recipient has previously discriminated against persons on the ground of race, color, or national origin, the recipient must take affirmative action to overcome the effects of prior discrimination. (Emphasis added.) 7 C.F.R. § 15.3(b)(6)(i).
Plaintiffs claim, and I agree, that the evidence clearly shows that the state’s essentially hands-off “freedom-of-choice” approach is demonstrably not “affirmative action to overcome the effects of prior discrimination,” within the contemplation of this specific regulation, and that neither does it comply with affirmative obligations imposed directly by the Fourteenth Amendment to disestablish formerly segregated state-supported facilities. Such action as the state has taken in this respect suffers the same vice as did comparable freedom-of-choice plans adopted in ostensible response to the Supreme Court’s mandate in Brown v. Board of Education, 347 U.S. 483, 74 S.Ct. 686, 98 L.Ed. 873 (1954); it has proven not to work — for the same obvious reasons that school freedom of choice plans did not work. See Green v. County School Board, 391 U.S. 430, 88 S.Ct. 1689, 20 L.Ed.2d 716 (1968). For the same reason that judicial remedies were available to insure the use of “other means” having more “realistic promise” to achieve the end compelled by law and constitution in school desegregation, so should a comparable judicial remedy be available here. See id. at 437, 88 S.Ct. at 1693.
The other prong of plaintiffs’ claim relates to single-race clubs formed since 1965 in communities where there demonstrably exists a potential for mixed-race .membership. As to those, the claim is that the state, through the Extension Service, intentionally countenances the continued organization and operation of these clubs, albeit by private parties, on a segregated basis, notwithstanding it has the means further to cause their desegregation.
That such clubs continue to exist is not disputed. That chance alone could account for their single-race composition is not suggested by anyone. The state’s position, essentially accepted by the district court, is that no invidious discrimination could be found as the cause except on the basis of evidence that specific black applicants had been denied membership in all-white clubs. Because that evidence was not forthcoming, the district court rejected any claim of present invidious discrimination.
I believe that the court erred in not addressing the question whether a pattern of invidious discrimination was nevertheless proven independently of specific rejection of black applicants for club membership. That necessary inquiry — which should be made first instance by the district court— was simply avoided by the court’s view that invidious discrimination could only be proven in the one way not proven. This clearly *699misapprehended controlling legal precedents for the means of proving invidious discrimination in this and related contexts. See, e.g., EEOC v. American National Bank, 652 F.2d at 1198-99 (discrimination in hiring by word-of-mouth recruitment by presently employed whites); Lea v. Cone Mills Corp., 301 F.Supp. 97, 102 (M.D.N.C. 1969), aff'd in relevant part, 438 F.2d 86 (4th Cir.1971) (discrimination in hiring by “chilling”).
I would reverse the district court’s judgment respecting the 4-H Club claim, and remand for further proceedings. As to those clubs which demonstrably existed on a state-imposed segregated basis prior to the effective date of Title VI16 I would direct entry of a remedial decree which required the state to “come forward with a new plan ... that promise[s] realistically,” see Green v. County School Board, 391 U.S. at 439, 88 S.Ct. at 1694, to work in achieving the fullest measure of desegregation of those clubs practically possible; with submission of that plan to the court required for review and approval as a condition to further receipt of public funds.17
As to those all-white clubs formed since the effective date of Title VI that now exist in communities with potential for mixed-race membership, I would direct specific reconsideration, on a re-opened record, of the claim that they are so maintained by invidiously discriminatory practices knowingly-countenanced and encouraged by the state through the Extension Service. In considering that claim, I would direct that evidence other than specific rejection of black applicants be considered and might suffice, if of sufficient probative force, to establish it. And I would direct that if such a claim were established, the court should enter a remedial decree comparable to that above described.
For these reasons, I dissent to the extent indicated.
. For sake of simplicity, I confine discussion of the salary claims to their grounding in Title VII and the Constitution. The Title VI ground, also available to the Government, must be essentially congruent, being rested on the prohibition against public funding of segregated facilities.
. In addition to the claim under Title VII, which only arose in 1972, the salary claim was based on Title VI, forbidding since 1965 discrimination in programs receiving federal funds, and on constitutional provisions proscribing intentional race discrimination by state action. As to the latter claims, a limitation period of three years applied to the action which commenced on November 18, 1971.
. The professional work-force then employed by the Extension Service obviously included both pre- and post-merger hires, a fact manifest to Blalock or anyone else knowledgeable as was he about the recent history of the Service.
. If, as plaintiffs plausibly contend, the average differentials should be adjusted by a tenure longevity factor, the gap-closing dates are substantially later, and may not have been closed fully by trial time.
. The district court’s rejection of Dr. Mann's regression analysis also, inevitably, involved rejection of Giesbrecht’s which concededly employed the same methodology and data bases. Supporting the district court’s position, the defendants are forced to the position — which seems clearly taken — of similarly repudiating their own expert witness testimony.
. The suggested variables were as follows:
(1) Performance of agents measured against the agents’ plan of work;
(2) The variations in salaries created by across the board state raises with the different percentage of state contributions in each county;
(3) The across the board increases in agent salaries by some counties and not in others;
(4) The merit raises provided by the state;
(5) The merit raises provided for by the counties in which Extension Service personnel have no input;
(6) The merit raises provided by the counties with limited or full participation in the merit recommendation by Extension Service personnel;
(7) The range in merit salary increases provided by the counties (0 to 12% in 1981);
(8) Prior and relevant experience; and
(9) Variations in salary due to market demands both at time of hire and later for agents with skills in short supply or prior experience.
. Indeed, one variable — merit—merely suggested by defendants as possible explanation for the differential was worked into the regression analyses. The result, however, was to increase the demonstrated race-effect, rather than decrease it. The Giesbrecht analysis for 1975 incorporated a quartile-ranking variable which presumably reflected merit evaluation. It had the effect indicated.
On this point, the majority opinion’s painstaking standard deviation analysis of the effects of quartile rankings on the plaintiffs’ claim, at 672-673, does no more than support the district court’s rejection of plaintiffs' discrete claim that the ranking system itself was violative of Title VII as an unvalidated test. I am prepared to accept it for that limited purpose without accepting the propriety of the majority’s first instance standard deviation analysis of statistical proof based on the size samples here involved, or the majority's apparent understanding of the import of the Hazelwood "more than two or three standard deviations" rule. See EEOC v. American National Bank, 652 F.2d at 1192-93 (“Within the range of one to three standard deviations ... we do not see how a court can properly find the ... hypothesis [of] discrimination dispelled by [the standard deviation] analysis alone.”).
To show, by whatever means, that the quartile ranking system did not itself violate Title VII does not however undercut the probative force of a regression analysis that did not use quartile rankings as an additional independent variable to test for race-effect. In short, that a discriminatory race-effect could not be traced solely to the quartile ranking system obviously does not also show that discrimination did not exist by virtue of other specifically identifiable causes or, indeed, unidentifiable causes. The majority, in fact, asserts that the race-effect demonstrated by all the regression analyses could be attributable only to the quartile ranking system or to the "lingering effects of pre-Act discrimination.” At 674. As will be shown, the fact that such lingering effects are traceable to pre-Act discrimination does not, as the majority believes, insulate it as a discriminatory post-act practice.
. Much of this evidence came directly from county-level Extension Service records.
. This evidence, incidentally, completely undercuts the defendants’ suggestion, apparently accepted by both the district court and the majority on this appeal, that the regression analyses were flawed because of different salary levels from county to county. The plaintiffs’ evidence indisputably established that the general pattern of average differentials existed in both "high-pay" and “low-pay" counties.
. Of course, conditions existing as of trial time would have been highly relevant to the question of the propriety and scope of declaratory and injunctive relief. But plaintiffs’ claims, both those of the private parties and of the Government, were of course not limited to prospective relief.
. Plaintiffs' claims were also grounded in the fourteenth amendment, whose effective date of course long predated the origins of the challenged salary pattern. While the majority does not address this point, I assume, for purposes of this discussion, that the Evans principle would apply as well to constitutional claims, so far as it had any application. But, in line with the ensuing discussion in text, I think Evans no more precludes proof of a post-limitation pattern of salary discrimination rooted in a pattern that originated in a pre-limitation period, than Hazelwood precludes proof of a post-Act pattern of salary discrimination that carries forward differentials originating in the pre-Act period.
*696For sake of simplicity, I discuss the point in terms alone of the ability to prove a Title VII claim of continuing violation by a pattern or practice that originated pre-Act.
. The majority also relies upon this theory to dismiss the probative force of the regression analyses showing statistically significant race effects in 1974 and 1975 salary structures. In the majority’s view, the fact that the data used in those analyses included pre-Act as well as post-Act hires flawed the analyses, since pre-Act hiring decisions could not properly be considered in assessing Title VII liability. At 672. With all respect, that the salary data included salaries paid after 1974 to pre-Act hires is beside the point. The claim here is of a post-Act discriminatory pattern or practice in salary payments, not in hiring or work-force composition. That some of the employees claiming post-Act discrimination in salary payments were hired (and paid discriminatory salaries) pre-Act, does not make irrelevant evidence that they, with others, were paid discriminatory salaries post-Act.
. Presumably, a wholly post-Act "pattern or practice” could only be found in specific post-Act employment decisions — either in the form of a general decision to continue a pre-Act pattern or practice as a matter of deliberate post-Act judgment, or in the form of a sufficient number of discrete decisions respecting individual employees to indicate a post-Act pattern or practice that merely continued pre-Act policies into the new period. These can be seen as "continuing” violations, in that they continue pre-Act discrimination rather than merely perpetuate "past effects,” see, e.g., Reed v. Lockheed Aircraft Corp., 613 F.2d 757 (9th Cir.1980).
. The court refused as well to certify a class of 4-H Club claimants on the same principal basis that it denied the claim on the merits: that no black member of the putative class had actually been denied membership in an all-white club. On this view, there literally was no class of persons who could make the claim the court identified. As later developed on my discussion of the merits of the claim, that simply misconceives the nature of plaintiffs' claim respecting the 4-H Clubs.
. The only claim of illegal discrimination in administration of the 4-H Program now pressed by plaintiffs relates to the continuation of these segregated local clubs. As plaintiffs themselves point out, the Extension Service has been able quite effectively to remove the vestiges of discriminatory practice in all other aspects of the Program, including, for example, the operation of summer camps. See note 17, infra.
. Though the composition of particular clubs has of course changed and varied in size over time, so that some difficulty might be encountered in establishing the continuity of a particular club from its origins in state-imposed segregation, the difficulty of proof is an evidentiary one no greater than that encountered in many other contexts. The defendants’ veiled suggestions that the difficulty makes a workable remedy impossible is disingenuous.
. The district court may well have been led into the error of outright rejection of this claim by assuming that only a much more Draconian remedy than the one suggested here as the appropriate one was possible. In another candid and revealing passage, the court opined in North Carolina as well as all other states integration of the races more frequently than not meets with strong resistance. The choice thus posed is whether it is better that the Extension Service continue to provide its much needed services to well over 100,000 North Carolina club members while striving to achieve full integration of the clubs or that it withdraw such services altogether as the government would have it do. The Extension Service has opted for the former, and in so doing this court does not perceive that it has violated the rights of anyone under any law. This misconceived both the nature of the plaintiffs’ claim and the remedial powers available to vindicate that claim if established. Within the equitable powers of the district court lay a much wider range of possible injunctive remedies than the immediate withdrawal of all state services to all 4-H Clubs found not presently desegregated to a specific degree. Neither Title VI nor the equal protection clause required a specific degree of desegregation of these clubs at any specific time. They required only that intentional segregation be neither directly effectuated nor symbiotically encouraged by state action and that state-imposed segregation still in place be rooted out by state remedial action to the maximum extent and as speedily as possible. As plaintiffs cogently observe, the district court’s assessment that no effort by the state beyond a benign freedom of choice plan could be required without risking destruction of the 4-H Clubs is belied by, among other things, the successful efforts by the Extension Service to desegregate all its other activities, including many involving 4-H Club members. The district court underestimated its powers to insure compliance by means less drastic and more adaptable to legitimate state interest than the horrible alternatives suggested by defendants as the only available ones.