with whom VAN DUSEN, Circuit Judge, joins, dissenting:
The majority opinion has wholly ignored the district court’s findings of fact which establish, beyond peradventure that the discharge of the plaintiff did not involve state action. Embarking on its own fact finding excursion, the majority of this court has sifted and weighed the evidence and has acted not as a reviewing authority but rather as an initial fact finder in seeking to support What I regard as an unwarranted conclusion — that state action exists. Primarily because the majority has transgressed its fundamental function as an appellate court, I am obliged to dissent.
That I differ somewhat from the majority in my reading of the applicable law is secondary in my view to the disregard shown by the majority for its proper function. It is clear to me as it should be to any appellate court that the facts found here by the district court, being fully supported by the evidence and therefore unassailable, make this a far different case than Hollen-baugh v. Carnegie Free Library, 545 F.2d 382 (3d. Cir. 1976). I would affirm the district court based upon its findings of fact which, being supported by evidence, cannot be deemed clearly erroneous, Krasnov v. Dinan, 465 F.2d 1298, 1302 (3d Cir. 1972)— the appropriate standard of review to which the majority has assiduously avoided addressing ¡itself.1
I.
After a four-day bench trial the district court found the following facts concerning *748the relationship of the City of Wilmington to the Wilmington Institute, and to Chal-fant’s firing in particular. So that the fact findings of the district court are fully set forth I have included the relevant footnotes to which the district court’s text refers.
In 1857 the “Young Men’s Association for Mutual Improvement of the City of Wilmington” received a corporate charter for the maintenance of a private subscription library (11 Del. Laws, Chap. 344); in 1859 the name of the private corporation was changed to “The Wilmington Institute” (11 Del. Laws, Chap. 516). After the passage of 19 Del. Laws, Chaps. 734, and 983 (1893), the Institute became a free public library “opened to the usé of citizens of Wilmington,” forfeiting its status as a private subscription library in exchange for an exemption from property taxes7 and annual appropriations from the City of five thousand dollars plus additional sums computed according to a formula based on the City’s population. A further provision of the -1893 law required that the Mayor and five other City officials were to become ex officio members of the Institute’s Board of Managers. However, the final section of this legislation expressly provided:
“That the managers of the said institute shall have power to make by-laws and rules for the government of the library and reading rooms.”
Twenty-eight years later8 in 1921, pursuant to yet another statute, the Institute agreed to transfer a tract of land having a value in excess of $200,000 located in downtown Wilmington to the City of Wilmington in exchange for $200,000, which it was then to combine with its privately endowed building fund of $300,000 in order to finance the construction of its main library. Under the terms of this statute (32 Del.Laws, Chap. 109), the Institute was authorized to retain possession of the land and main library building so long as a free public library was maintained on the premises.9 The library property was finally deeded in fee to the City on March 23, 1923 and the City in return granted a perpetual lease of the building back to the Institute with the right of management being held by the Institute. (Docket Item 31, par. 2). At about the same time the City also by indenture recognized that ownership of all books and documents housed in the library belonged to the Institute. (Docket Item 31, par. 29).
While the Institute’s library service obligations to the public expanded between 1920 and the early 1970’s through the influx of greater amounts of public funding, first from the City and then from the County, the active management of the Institute’s operations was retained by the self-perpetuating Board of Managers, a majority of whom were and are private citizens.13 Henry R. Folsom, Jr., County Councilman from 1967 until 1973 and President of County Council at the time of trial (Tr. 288-9), testified that during these years at “budget time,” the County Council would invariably ask the Institute for some explanation as to whether the funds slated for appropriation were to be used for the purchase of books or for salaries. (Tr. 292, 303). He also testified that the County Council did not attempt to control the expenditure of the funds thus appropriated (Tr. 301-2),14 and *749emphatically denied that the New Castle County government was involved or even concerned about the daily operation of the Institute, including the hiring, firing or promotion of employees and administrators. (Tr. 290, 301, 302). County appropriations for calendar year 1972 were $416,000 (Docket Item 31, par. 6(c)) and City appropriations for fiscal year 1971-1972 totaled $537,000 (PX 41), but other than these large sums of money routinely granted to the Institute there is no other evidence of external governmental supervision over the Institute’s Board of Managers or its Director. There is also a lack of evidence concerning the involvement of the ex officio members of the Board of Managers who are City or County officials (Tr. 732, 677-8), and it is clear that the convening of the Special Committee [to reconsider Chalfant’s discharge], comprised of three non-governmental Board members, was primarily, if not exclusively, the decision of duPont and Abrams. (Tr. 679-680, 695, 733; Docket Item 35 at p. 3).
417 F.Supp. 160, 163-65 (D.Del.1976) (some footnotes omitted).
As an appellate tribunal, we are bound by the district court’s findings of fact unless they are “clearly erroneous.” Fed.R.Civ.P. 52(a); Krasnov v. Dinan, supra, 465 F.2d at 1302. Each of Judge Latchum’s findings of fact is fully substantiated by evidence in the record. His findings are also clearly articulated so that we as an appellate court may properly evaluate each of them and apply the appropriate standard of review. See O’Neill v. United States, 411 F.2d 139, 146 (3d Cir. 1969); Bowles v. Cudahy Packing Co., 154 F.2d 891, 894 (3d Cir. 1945). Chalfant therefore does not, and could not, contend that Judge Latchum’s findings are clearly erroneous: indeed, at oral argument Chalfant conceded that she did not challenge them. Even Judge Gibbons, writing for the majority, acknowledges that “[m]ost of the historical or narrative facts upon which the [District] Court predicated its legal conclusion that state action was lacking are not disputed in any material respect.” Maj. Op. at 742.
Yet the majority opinion, despite its recognition that the district court’s findings of fact are not clearly erroneous, wholly disregards those findings. Instead the majority asserts: “There is additional, undisputed evidence in. the record which is relevant to the governmental status of the Wilmington Institute, but which the District Court, in its opinion below, did not specifically mention.” Id. at 742. The majority opinion then proceeds to winnow from the record selective pieces of evidence which it finds will support the result that state action exists.
For example, the majority discusses at length two funding and record-keeping agreements between New Castle County and the Wilmington Institute (PX 36 and PX 37). Maj. Op. at 742-743. Although these agreements were not in- effect at the time of Chalfant’s discharge in 1972, the majority opinion relies upon them to make its own unsubstantiated finding that “they do show the nature of the relationship between the county government and the public library, a relationship which antedated federal funding.” Id. at 743.2 By contrast, with that same evidence before it, the district court *750found that “at least until 1974, neither the City of Wilmington nor New Castle County probed into the Institute’s budgetary needs or fiscal administration, or meddled in its overall management, let alone its employment practices.” 417 F.Supp. at 165-66.
Similarly, to demonstrate that the Institute was government operated and not privately controlled, the majority quotes at length from a Wilmington Institute Staff Handbook, emphasizing that “[t]he members of the Board of Managers donate their time and energies to the library as a community service. The Director and all staff members are paid public servants ” (PX 35). Maj. Op. at 741 (emphasis in majority opinion).3 This handbook was in evidence and before the district court. Yet the district court, contrary to what the majority would find, found on this and other evidence that the significant factor concerning library personnel was that the Institute’s “active management” was “retained” by the “Board of Managers, a majority of whom were and are private citizens,” 417 F.Supp. at 165, and that apart from government funding “there is no other evidence of external governmental supervision over the Institute’s Board of Managers or its Director.” Id.4
Fairly read, the majority opinion stands for the proposition that, despite findings of fact made by the district court which cannot be overturned because they are supported by evidence, the presence in the record of additional evidence which has not been specifically mentioned by the district court permits a court of appeals to find its own facts, without being bound by the district court’s findings. I find that proposition completely unique and unsupportable. There is no precedent for such a standard of review. To the contrary, this court and others have consistently held that a district court is not required to make findings on all the evidence presented if the findings that it does make are sufficient to support its ultimate conclusion. Furthermore, the district court need not make findings which assert the reciprocal negative of each of its affirmative findings. Rayonier Inc. v. Poison, 400 F.2d 909, 923 (9th Cir. 1968); see Bowles v. Cudahy Packing Co., supra, 154 F.2d at 894. Having found for example that Chalfant had been discharged on February 16, 1972, the district court was not required to find that she was not discharged on February 15th or February 17th, or on any other date in any other year. Accordingly, having found that the evidence (particularly the testimony of Edward duPont, President of the Board of Managers) established control of the Insti*751tute by “private citizens,”5 Judge Latchum was not required to “find” that the one piece of conflicting evidence (the letter from William J. Conner (PX 19)), did not suffice to establish governmental control. Similarly, having found that the evidence established that the government was not involved with the Institute’s budgetary management and employment policies before 1974, Judge Latchum was under no duty to add to that finding that the post-discharge agreements (PX 36 & PX 37) were insufficient to establish that the government was involved before 1974.
By rejecting the district court’s evaluation and assessment of the evidence, however, the majority opinion has now apparently imposed a duty of making reciprocal findings upon the district court. The majority's rationale would require in effect that the district court negate each of its affirmative findings by a separate, independent reciprocal finding which recites the unpersuasiveness of the evidence which it rejects.
The majority having determined that the district court did not meet this novel standard which is now to be employed, one would then expect that the majority would remand the proceedings to the district court for that purpose. Not so. Rather, the majority has substituted its own process of evaluating and assessing certain isolated pieces of evidence in order to arrive at a legal conclusion compatible with the result it prefers. This conclusion, as I have stated, has been reached wholly without reference to the actual findings made by the district court — findings by which we are bound.
Our court has previously stated:
The trial court is not required to make findings on all the facts presented . . . The judgment should stand if the opinion below gives the appellate court a clear understanding of the basis of the decision.
Bowles v. Cudahy Packing Co., supra, 154 F.2d at 894. More recently our court added to this “sufficiency” standard the requirement that the district court make subordinate factual findings in support of its ultimate findings so as to allow us to ascertain “what evidence the District Judge accepted as credible or what he rejected,” O’Neill v. United States, supra, 411 F.2d at 146, and to provide “a basis for review of the District Judge’s finding[s],” id. Accord, Kreda v. Rush, 550 F.2d 888, 890 (3d Cir. 1977) (case remanded to the district court “to specify the reasons on which its disposition is premised so as to inform the parties and facilitate review”); FTC v. British Oxygen Company, Ltd., 529 F.2d 196, 200 (3d Cir. 1976) (en banc) (remand ordered as “there is no means for the reviewing court to determine what evidence served as the basis for the injunction”); Pepi, Inc. v. Helcar Corp., 458 F.2d 1062, 1064-65 (3d Cir. 1972) (quoting O’Neill).
In sharp contrast to the findings made in O’Neill, Kreda, British Oxygen, and Pepi, the district court here met the sufficiency standard by making extensive subordinate and ultimate findings. By so doing the district court has afforded us a proper basis for review. See Kreda v. Rush, supra, 550 F.2d at 890; FTC v. British Oxygen Co., Ltd., supra, 529 F.2d at 200; O’Neill v. United States, supra, 411 F.2d at 146.
Other courts of appeals have adopted similar standards of sufficiency. The Court of Appeals for the Ninth Circuit explained:
the federal rule relating to findings of a trial court does not require the court to make findings on all facts presented or to make detailed evidentiary findings; if the findings are sufficient to support the ultimate conclusion of the court they are sufficient. [Citation] Nor is it necessary that the trial court make findings asserting the negative of each issue of fact raised. It is sufficient if the special affirmative facts found by the court, construed as a whole, negative each rejected contention. The ultimate test as to the adequacy of findings will always be whether they are sufficiently comprehensive and pertinent to the issues to provide *752a basis for decision and whether they are supported by the evidence.
Rayonier Inc. v. Poison, supra, 400 F.2d at 923, quoting Carr v. Yokohama Specie Bank, Ltd., 200 F.2d 251, 255 (9th Cir. 1952); accord, Amerline Corp. v. Cosmo Plastics Co., 407 F.2d 666, 669-70 (7th Cir. 1969) (district court “was not required to make findings with respect to all evidence introduced”); Weaver v. United States, 334 F.2d 319, 320-21 (10th Cir. 1964); see 5A J. Moore, Moore’s Federal Practice 152.06[1], at 2710, 2713-16 & n.26 (2d ed. 1977) (“the court need not find [facts] on every issue requested, but a finding of such essential facts as lay a basis for the decision is sufficient”; “[findings need not assert the negative of rejected propositions”), citing Committee Note of 1946 to Rule 52(a), reprinted in 5A Moore’s Federal Practice, supra, H 52.-01[6], at 2606-07 (“the judge need only make brief, definite, pertinent findings and conclusions upon the contested matters; there is no necessity for over-elaboration of detail or particularization of facts”); 9 C. Wright & A. Miller, Federal Practice and Procedure § 2579, at 715 (1971) (“[t]he ultimate test of the adequacy of a trial judge’s findings is whether they are sufficiently comprehensive and pertinent to the issues to provide a basis for decision”).
The First Circuit Court of Appeals has articulated this rule even more forcefully:
It is settled that an appellate court has no power to disturb a finding of fact of a trial court where it is based on some substantial though conflicting evidence. Jue v. Bass, 299 F.2d 374, 377 (9th Cir. 1962). An appellate court, in determining the presence or absence of substantial evidence, is not required to reweigh the evidence adduced in the trial court but need only scrutinize the .record to ascertain if it affords some reasonable basis for the result achieved. Lattig v. Pilliod, 289 F.2d 478 (7th Cir., 1961).
“It is not our function to hold a trial de novo in which the whole controversy begins anew and in which every disputed factual controversy and every possible inference is earnestly debated, measured, evaluated and tallied only to reach for the next disputed factual issue to inventory in like fashion.” Travelers Insurance Company v. Truitt, 280 F.2d 784, 788 (5th Cir., 1960).
In sum, we are not to substitute our judgment for that of the trial court, Homestake Mining Co. v. Mid-Continent Exploration Co., 282 F.2d 787, 796 (10th Cir., 1960), but may reverse only where, upon a consideration of the entire evidence, we are “left with a definite and firm conviction that a mistake has been committed.” McAllister v. United States, 348 U.S. 19, 20, 75 S.Ct. 6, 99 L.Ed. 20 (1954).
Evans v. United States, 319 F.2d 751, 753 (1st Cir. 1963).
The majority opinion here does precisely what our sister circuits have renounced, for the majority has “h[e]ld a trial de novo in which . . . every disputed factual controversy and every possible inference is earnestly debated, measured, evaluated, and tallied.” This role is of course for the fact-finder, not for us. Precisely because reasonable minds differ as to the probative value and weight of evidence,6 it is the district court judge who, having conducted the trial, is accorded the role of fact-finder. While we might well assess the facts differently were we the district court, we are not. Furthermore, not being the fact-finder, we may not overturn findings on appeal unless they are clearly erroneous. Judge Aldisert has stated this principle in such unequivocal and classic terms that it bears repetition here:
“In reviewing the decision of the District Court, our responsibility is not to substitute findings we could have made had we been the fact-finding tribunal; our sole function is to review the record to determine whether the findings of the District Court were clearly erroneous, i. e., whether we are ‘left with a definite and firm conviction that a mistake has been committed.’ ” Speyer, Inc. v. Hum*753ble Oil and Refining Co., 403 F.2d 766, 770 (3d Cir. 1968). It is the responsibility of an appellate court to accept the ultimate factual determination of the fact-finder unless that determination either (1) is completely devoid of minimum evi-dentiary support displaying some hue of credibility, or (2) bears no rational relationship to the supportive evidentiary data. Unless the reviewing court establishes the existence of either of these factors, it may not alter the facts found by the trial court. To hold otherwise would be to permit a substitution by the reviewing court of its finding for that of the trial court, and there is no existing authority for this in the federal judicial system, either by American common law tradition or by rule and statute.
Krasnov v. Dinan, supra, 465 F.2d at 1302.
In ignoring these fundamental precepts, Judge Gibbons, writing for the majority, is guilty of the very process which he condemned in his dissenting opinion in Lin-mark Associates, Inc. v. Township of Will-ingboro, 535 F.2d 786, 805 (3d Cir. 1976), rev’d, 431 U.S. 85, 97 S.Ct. 1614, 52 L.Ed.2d 155 (1977), when he stated in reference to the Linmark majority:
in utter disregard of the standard of review imposed on this court by Rule 52, Fed.R.Civ.P., the majority, by a disingenuous process of selection and omission, sifts through the record below to construct its own findings to justify a predetermined result. . . . [Its] opinion completely inverts the respective roles of the trial and appellate courts, and is an instance of ad hoc decision making.
Id. at 810.7
Having failed to demonstrate that the district court’s facts were not supported by evidence, this court is necessarily bound by the district court’s findings and cannot avoid the ultimate conclusion to which they lead by usurping the district court’s function and finding its own facts.
II.
If the district court’s findings of fact are accepted, as I believe they must be on the present record, then the result reached by the majority opinion cannot stand.
The following findings are particularly instructive to our legal analysis and bear emphasis.
1. The Institute was governed by a Board of Managers which consisted of 10 nongovernmental members and 8 ex officio members. 417 F.Supp. at 165 n.13.
2. “[T]he active management of the Institute’s operations was retained by the self-perpetuating Board of Managers, a majority of whom were and are private citizens.” Id. at 165.
3. During 1972 and 1973, approximately 90 per cent of the Institute’s income was contributed by the City of Wilmington and New Castle County. Id. at 165 & n.14; App. at 28-30.
4. “[O]ther than [the] large sums of money routinely granted to the Institute there is no other evidence of external govérnmental supervision over the Institute’s Board of Managers or its Director.” Id. at 165.
5. “There is also a lack of evidence concerning the involvement of the ex officio members of the Board of Man*754agers who are City or County officials . . . 8 Id. at 165.
6. “[I]t is clear here that at least until 1974, neither the City of Wilmington nor New Castle County probed into the Institute’s budgetary needs or fiscal administration, or meddled in its overall management, let alone its employment practices.” Id. at 165-66.
7. “The record conclusively shows that the State of Delaware has never had any contention with the funding or operation of the Institute.” Id. at 163 n.6.
(Emphasis added.)
As this case is being considered en banc, we look first to most recent Supreme Court decisions to determine whether the discharge of Chalfant by the Wilmington Institute constituted state action. Three categories of state action defined by the Supreme Court and discussed in recent opinions of this court are:
(1) where state courts enforced an agreement affecting private parties [e. g., Shelley v. Kraemer, 334 U.S. 1, 68 S.Ct. 836, 92 L.Ed. 1161 (1948)]; (2) where the state “significantly” involved itself with the private party [e. g., Burton v. Wilmington Parking Authority, 365 U.S. 715, 81 S.Ct. 856, 6 L.Ed.2d 45 (1961)]; and (3) where there was private performance of a government function [e. g., Terry v. Adams, 345 U.S. 461, 73 S.Ct. 809, 97 L.Ed. 1152 (1953) (white primary case)].
Hollenbaugh v. Carnegie Free Library, 545 F.2d 382, 383 (3d Cir. 1976), quoting Magill v. Avonworth Baseball Conference, 516 F.2d 1328, 1331 (3d Cir. 1975).
. It is obvious that Chalfant does not fall into the first category. This is clearly not a case in which “the state courts enforced an agreement affecting private parties.” Nor do I believe that Chalfant involves the third category: “the exercise by a private entity of powers traditionally exclusively reserved to the State” or “traditionally associated with sovereignty.” Jackson v. Metropolitan Edison Co., 419 U.S. 345, 352-53, 95 S.Ct. 449, 42 L.Ed.2d 477 (1975). Whereas the majority opinion implicitly attempts to bring this case within that framework,9 I believe that our court was correct when it stated but one year ago that “nor can it be said that the operation of a library constitutes private performance of a function traditionally associated with sovereignty.” Hollenbaugh, supra, 545 F.2d at 383.10
In contending that state action is present in this case, Chalfant has concentrated on the second category identified in Hollen-baugh. The crux of her argument, and of *755the majority’s opinion, is that here we are presented with a case in which the state is “significantly” involved with a private par-. ty. To test that thesis requires a two-fold inquiry:
“whether there is a sufficiently close nexus between the State and the challenged action ... so that the action of the latter may be fairly treated as that of the State itself,” Jackson v. Metropolitan Edison Co., 419 U.S. 345, 351, 95 S.Ct. 449, 453, 42 L.Ed.2d 477 (1974), citing Moose Lodge No. 107 v. Irvis, 407 U.S. 163, 176, 92 S.Ct. 1965, 32 L.Ed.2d 627 (1972), or whether the “State has so far insinuated itself into a position of interdependence with [the library] that it must be recognized as a joint participant in the challenged activity.” Burton v. Wilmington Parking Authority, 365 U.S. 715, 725, 81 S.Ct. 856, 862, 6 L.Ed.2d 45 (1961).
Hollenbaugh, supra, 545 F.2d at 383 (brackets in original).
Contrary to the majority, I am convinced that the findings of fact made by the district court in this case, each of which is supported by evidence, fail to demonstrate either a “nexus” or an “interdependence” sufficient to establish state action by the Institute.
I reiterate the position expressed in my concurrence in Braden v. University of Pittsburgh, 552 F.2d 948, 965 (3d Cir. 1977) (en banc) that the Burton “symbiosis” standard has been strictly narrowed if not eroded by subsequent Supreme Court precedent. 552 F.2d at 973. I explained in Braden that
Moose Lodge, Gilmore [v. City of Montgomery, 417 U.S. 556, 573, 94 S.Ct. 2416, 41 L.Ed.2d 304 (1974)], and Jackson place a gloss upon Burton which we are not free to ignore. The teaching of those cases is that “neither general government involvement nor even extensive detailed state regulation is sufficient for a finding of state action. Rather, the state must affirmatively support and be directly involved in the specific conduct which is being challenged.” (Footnote omitted.) Cannon v. University of Chicago, [559 F.2d 1063, 1069 (7th Cir. 1976), aff’d on rehearing, 559 F.2d 1077 (7th Cir. 1977) (Title IX claim)].
552 F.2d at 973.11
Assuming however that the majority is correct in its interpretation of Burton and that state action can be demonstrated absent governmental involvement in the specific challenged act, I am still unable to find a Burton -type symbiosis here.
In Burton the Supreme Court “held that a private restaurant owner who refused service to a customer because of his race, violated the Fourteenth Amendment where the restaurant was located in a building owned by a state-created parking authority and leased from that authority.” Braden v. University of Pittsburgh, supra, 552 F.2d at 957. The present case is not comparable to Burton for at least three reasons.
First, Burton, unlike this case, involved racial discrimination, the principal evil against which the fourteenth amendment was directed. As our court recently recognized, state action precedents which developed in racial discrimination cases do not have the same force in other contexts.12
Second, in Burton, unlike this case, the appellate court was not faced with detailed findings of fact showing that the relevant governmental units actually exerted no control over the actions of the private entity. In Burton the trial court entered judgment for the plaintiff after all the parties had conceded that there were no material facts in dispute. Burton v. Wilmington Parking Authority, 38 Del.Ch. 266, 267, 150 A.2d 197, 198 (Ch.1959). In Burton, no attempt was made by the defendant parking authority to show a lack of control by the authority over the restaurant. Here, on the other hand, evidence was introduced from which the district court found that state played no *756role in the challenged actions and that it did not control the operations or management of the Institute in general. As a result, this appeal stands on an entirely different footing from the appeal in Burton.
Finally, Chalfant has failed to prove that the relationship between the state and the Institute was “symbiotic”13 or one of “interdependence” 14 in the same sense as the relationship between the restaurant and the parking authority in Burton. In Burton, the rent received by the parking authority from the lease of portions of its parking garage was an “indispensable part of the State’s plan to operate its project as a self-sustaining unit.”15 365 U.S. at 724, 81 S.Ct. at 861. The restaurant drew additional customers because of its convenient location, and it alleged that “for it to serve Negroes would injure its business.” Id. Consequently, the Supreme Court concluded that “profits earned by discrimination not only contribute to, but also are indispensable elements in, the financial success of a governmental agency.” Id. at 724, 81 S.Ct. at 861. In other words, the restaurant earned extra profits because of its location in a public facility, and the State received “profits earned by discrimination,” a practice in which it could not itself engage. Thus the relationship between the State and the restaurant in Burton was a classic illustration of symbiosis: a relationship in which each partner does for the other something which the other partner needs but cannot do for itself.
In this case, by contrast, no such relationship was proved. The district court found:
Plaintiff has failed to show that this public funding was in any way motivated by the City’s and County’s belief that it would be less expensive to support a privately run library, which might be able to operate on a lower salary scale and with fewer employee benefits than either governmental body would be required to spend to operate a library or libraries directly, i. e., on their own and in their own facilities.
417 F.Supp. at 166.
Undoubtedly the state benefited in a sense from the services provided by the Wilmington Institute, since the Institute’s facilities are open to the people of the Wilmington area. But there is no evidence to support a finding that the state was “dependent” upon the Institute any more than any government is “dependent” upon those innumerable charitable and nonprofit organizations which it exempts from taxes.
Nor do the district court’s findings reveal that the state significantly participated in the Institute’s day-to-day or overall affairs —financial, managerial or otherwise. See id. at 163-65. The majority opinion “concede^]” that “the city has no interest in the tenant’s revenues.” Maj. Op. at 745. And as previously stated the district court found no supervision exercised by the state over the Institute. Hence the majority’s attempt to force this case “squarely within even the narrowest reading of Burton ”, id., must necessarily fail.
Nor can Chalfant prevail under Jackson v. Metropolitan Edison Co., 419 U.S. 345, 95 S.Ct. 449, 42 L.Ed.2d 477 (1974), Gilmore v. City of Montgomery, 417 U.S. 556, 94 S.Ct. 2416, 41 L.Ed.2d 304 (1974), or Moose Lodge *757No. 107 v. Irvis, 407 U.S. 163, 92 S.Ct. 1965, 32 L.Ed.2d 627 (1972), because she failed to establish a “nexus” between the state and the challenged actions of the Institute and its officers. See 419 U.S. at 351, 95 S.Ct. 449 ; 417 U.S. at 573, 94 S.Ct. 2416; 407 U.S. at 176-77, 92 S.Ct. 1965. Indeed, Chal-fant conceded at oral argument that the state was not involved in her dismissal and that therefore she could not meet the Jackson test.
There was no evidence to sustain a finding that the state played any role in Director Bryant’s initial decision to dismiss Chalfant. When Chalfant’s attorney requested reconsideration of that decision, he wrote to Bryant and to Edward duPont and Richard Abrams, two non-governmental members of the Board. The district court found that “the convening of the Special Committee, [which ratified Bryant’s decision,] was primarily, if not exclusively the decision of duPont and Abrams.” 417 F.Supp. at 165. The hearing conducted by the Special Committee was held on private premises with duPont presiding. There is no evidence and hence there could be no finding, that the entire Board was asked to, or sought to, review the Special Committee’s decision. There is also no evidence to suggest that government officials or entities had anything to do with a letter concerning Chalfant which duPont wrote to the American Library Association (ALA). It therefore seems clear, and Chalfant agrees, that there is no nexus between the state and the actions which form the basis of Chalfant’s complaint. Consequently, there is no state action that can be said to exist under the theory of Jackson v. Metropolitan Edison Co., Gilmore v. City of Montgomery, and Moose Lodge No. 107 v. Irvis.16
In sum, based upon the district court’s findings of fact, it is evident Chalfant has failed to demonstrate either a “symbioftic]” relationship between the state and the Institute or a “nexus” between the state and Chalfant’s discharge.
III.
I also disagree with the majority in its reading of this Circuit’s recent precedent on the state action question. Specifically, I cannot agree that “[c]ompared to Hollenbaugh [where we found state action], this is an a fortiori case.” Maj. Op. at 745.
In Hollenbaugh,17 discharged library employees brought suit against their employer under 42 U.S.C. § 1983. The district court granted summary judgment in favor of the library based upon a perceived lack of state action. This Court reversed, holding that “[t]he totality of circumstances compels our conclusion that state involvement in the library’s operation was significant.” 545 F.2d at 385.
*758In its analysis, the Hollenbaugh court determined that the “ ‘receipt of money from the State is not, without a good deal more, enough to make the recipient an agency or instrumentality of the Government.’ ” Id., quoting Grossner v. Trustees of Columbia University, 287 F.Supp. 535, 547-48 (S.D.N.Y.1968). But, in Hollenbaugh the court found that “there was a good deal more.” 545 F.2d at 385. The “good deal more” in Hollenbaugh consisted not only of 90% government financing, but also, a tax expressly designed to support the library had been imposed; there was an explicit representation that the library was to act as agent for and on behalf of the City; and the governmental authorities had the power to appoint a majority of the library’s board of trustees. Id.18 The precise language used by the Hollenbaugh court and which led to its conclusion that state involvement in the library’s operation was significant reads:
But here we believe there was a good deal more. There was massive government financing, in the vicinity of 90 percent rather than the 49 or 51 percent discussed in Poindexter [v. Louisiana Financial Assistance Comm’n, 275 F.Supp. 833, 854 (E.D.La.1967), aff’d mem., 389 U.S. 571, 88 S.Ct. 693, 19 L.Ed.2d 780 (1968)]; and the city had authorized the imposition of a tax for the express purpose of supporting the library. Although concededly not controlling, we believe these are significant factors. Unlike Ma-gill, where “[plaintiff’s evidence did not establish that the program was represented as being sponsored by the municipalities”, 516 F.2d at 1335, here there was an explicit representation that the library was to act “as agent for and on behalf of” the school district and the city in providing public library service. In addition, a majority of the library’s trustees were appointable by governmental bodies. The totality of these circumstances compels our conclusion that state involvement in the library’s operation was significant.
545 F.2d at 385.
Here, however, with the exception of “government financing,” a significant but not controlling factor in the Hollenbaugh calculus, none of the other Hollenbaugh factors is present.19 Whereas in Hollen-baugh an express tax had been imposed and an explicit representation of agency had been made, here no such factors can be found.20 Moreover, whereas in Hollen-baugh the majority of the library’s board of *759trustees were appointed by governmental bodies, here by contrast a majority of the Institute’s Board of Managers was privately appointed.21 Hence, except for the receipt of money from units of local government, a factor expressly rejected as constituting state action without more,22 none of the factors present in Hollenbaugh is present here. Consequently in this case, there is little evidence of significant governmental involvement with the Institute. Indeed, the district court found after trial that the “active management” of the Institute was controlled by the privately-appointed majority of the Board of Managers and that “other than [the] large sums of money routinely granted to the Institute there is no other evidence of external governmental supervision over the Institute’s Board of Managers or its Directors.” 417 F.Supp. at 165.23 Thus, applying the “totality of circumstances” test stated in Hollenbaugh,24 *760the Wilmington Institute cannot be characterized as a state entity for purposes of § 1983.
Essentially therefore I disagree with the majority because of the importance of the district court’s fact findings which the majority has overlooked. It is for that reason that this case is far different from other state action cases and in particular from Hollenbaugh. Were it not for the explicit and detailed findings of fact made by the district court judge which compel the conclusion that no state action exists, this dissent would not have been necessary. But giving proper credit to the district court’s fact findings in the first instance, and then having analyzed state action doctrine as applied to those findings, I would affirm the district court’s judgment in favor of the Institute.
IV.
I would also affirm the district court’s dismissal of Chalfant’s pendent state claims. The district court based the dismissal of Chalfant’s pendent claims on alternative grounds: (1) that, under United Mine Workers v. Gibbs, 383 U.S. 715, 725, 86 S.Ct. 1130, 16 L.Ed.2d 218 (1966), it lacked power to hear those claims since they did not concern the same “nucleus of operative fact” as her federal claims; and (2) that, even if the court had the constitutional power to hear those claims, it would decline to do so in the exercise of its discretion.
There are strong grounds for believing that Chalfant’s federal claims and her pendent state claims do not share the same “nucleus of operative fact.” See PAAC v. Rizzo, 502 F.2d 306, 312-13 (3d Cir. 1974). duPont’s letter to the ALA, which forms the basis of Chalfant’s state defamation claims (Count II of her Amended and Supplemental Complaint), is never mentioned in the detailed factual allegations pertaining to her federal claims (Count I). In
addition, duPont’s letter to the ALA was not written until almost two years after Chalfant’s dismissal.
Nevertheless, I would find it unnecessary to determine whether the district court had the power to determine Chalfant’s pendent state claims, since I am convinced that it did not abuse its discretion in declining to do so. Gibbs states:
It has consistently been recognized that pendent jurisdiction is a doctrine of discretion, not of plaintiff’s right. . Needless decisions of state law should be avoided both as a matter of comity and to promote justice between the parties, by procuring for them a surer-footed reading of applicable law. . . . Similarly, if it appears that the state issues substantially predominate, whether in terms of proof, of the scope of the issues raised, or of the comprehensiveness of the remedy sought, the state claims may be dismissed without prejudice and left for resolution to state tribunals.
383 U.S. at 726-27, 86 S.Ct. at 1139. Gibbs also counsels that the district court retains the discretionary authority to dismiss pendent state claims “throughout the litigation.” Id. at 727, 86 S.Ct. 1130. See also PAAC v. Rizzo, supra, 502 F.2d at 312-13.
In this case, the district court recognized that the resolution of Chalfant’s state defamation claims might require it to decide numerous unsettled questions of Delaware law. In defense to Chalfant’s defamation claims, the defendants asserted several qualified privileges which have been accepted in other jurisdictions but which either have not yet been recognized or are not clearly delineated under Delaware law. 417 F.Supp. at 169; see Answering Post-Trial Brief of Defendants at 53-59. Given “the unsettled nature of state law” on these questions25 and “[i]n light of the broad discretion which district courts must be giv*761en in evaluating such matters,”26 I believe that the district court did not abuse its discretion in dismissing Chalfant’s pendent state claims.27
In all respects therefore, I would affirm the judgment of the district court.
. While the majority opinion recites that it “incorporates” the district court’s fact finding in its discussion, Maj. Op. at 742. it appears strange to me that the very foundation of the district court’s decision — its detailed fact finding — is not repeated in haec verba by the majority. This omission becomes even more curious when the majority opinion reprints verbatim a portion of the pretrial order, Maj. Op. at 741, but completely ignores the trial findings which are a product of all the evidence, including the facts admitted in the pretrial order. To supply this deficiency I have taken the liberty of setting forth the district court’s findings of fact in the text of this dissent.
This exemption is today codified in 9 Del.C. ■ § 8104, and is enjoyed by over seventy-five other organizations.
Between 1893 and 1921 the following legislation pertaining to the Institute was enacted. 22 Del.Laws, Chap. 360 (1905) authorized the City of Wilmington to contract with the Institute to provide unlimited annual appropriations; 25 Del.Laws, Chap. 105 (1909) raised the minimum annual appropriation which the City was required to provide the Institute.
Companion legislation (32 Del.Laws, Chap. 110) expanded the number of City officials who were to serve ex officio on the Board of Managers from 5 to 8 specifying, however, that 10 other persons were to be elected to the Board of Managers by the stockholders of the Institute.
The Board of Managers consists of ten self-perpetuating non-governmental members and eight ex officio governmental members.
Beginning with fiscal year 1973 (July 1, 1973-June 30, 1974), the Institute and New Castle County entered into a formal written agreement (PX 36) breaking down into nine discrete categories the uses to which the money appropriated b the County was to be put. The agreement refers to the fact that the County was slated to receive federal “entitlement funds” under the State and Local Fiscal Assistance Act'of 1972, 31 U.S.C. § 1221 et seq. These federal funds, to the extent received up to $481,581, were to be paid to the Institute for the operation of the two County libraries located outside of Wilmington. The Court is of the opinion that this formalization of the Institute’s status as the operator of the County’s library system was prompted by administrative provisions of federal law, 31 U.S.C. §§ 1241-1243, and in no way rendered dubious Folsom’s uncontradicted testimony (Tr. 290) concerning the period between 1967 and 1973, the years in question.
. The majority opinion fails to explain how the funding and record-keeping agreements which post-dated Chalfant’s discharge can have any effect upon the status of the institute at a time prior to their execution.
. If I were the district court fact-finder rather than a member of a reviewing tribunal, I would interpret that portion of the handbook not as indicating a sovereign (“public”) function as distinct from a private function but rather as contrasting the voluntary nature of the Board’s duties with the salaried status of other personnel. However, as I have attempted to demonstrate throughout this dissent, those facts which we as appellate judges might find based I upon our varying interpretations of the evidence are completely irrelevant. Our only concern can be with whether the evaluation of evidence and the findings of fact derived therefrom by the district court judge are supported by evidence in the record, regardless of our agreement or disagreement with the facts so found.
. Two additional examples are noted because of the importance given them in the majority opinion. First the majority refers to the county government’s interest in personnel policies, as demonstrated by reference to a letter in evidence written by the county executive (PX 19) and reproduced on page 743 of the majority opinion. The impression is thereby given that the district court’s finding of fact that “the active management of the Institute’s operations was retained by the self-perpetuating Board of Managers, a majority of whom were and are private citizens,” 417 F.Supp. at 165, is clearly erroneous. Yet the majority does not and cannot so hold — a flaw which I find fatal to the majority’s position.
Again, the majority opinion sets forth an interrogatory which was answered by the Institute, admitting that the Institute “is not a corporate entity [but] is a management title under which City and County libraries are operated.” (Docket Item 31). Maj.Op. at 743. Significantly neither the interrogatory nor its answer was in evidence, but even had they been, this isolated admission, and the county executive’s letter referred to by the majority, could constitute no more than two fragments of a congery of evidence from which the district court ultimately made its findings.
. 417 F.Supp. at 165.
. See, e. g., nn.2-4 supra.
. The majority opinion has apparently announced a new and unique standard of review applicable to findings of fact. That standard is couched in terms of: do “[a]ll of these facts and circumstances convince us”? Maj. Op. at 745. As I understand our procedure in this Circuit, an en banc court may overrule prior precedent whereas a panel is precluded from doing so. Internal Operating Procedures of the United States Court of Appeals for the Third Circuit, § M.2 (1974). However, even if the majority here intended adopting a standard of review different from the "clearly erroneous” standard recited in Krasnov v. Dinan, supra, it would be precluded from doing so by Fed.R. Civ.P. 52(a) which prescribes this precise standard. However, I do not attribute such an intention to the majority. I question only the majority’s refusal to treat fairly with the correct standard and the majority’s willingness to adopt a substantially different and less rigorous test.
. The district court clearly believed the following testimony of duPont concerning the manner in which the Board of Managers functions (see 417 F.Supp. at 165):
Q. . Can you tell me who elects the Board of Managers?
A. The Board of Managers — we have a nominating committee which we appoint, and in effect we elect ourselves. The stock is held in trust, or I believe it is the Bank of Delaware, and we in effect elect ourselves.
Q. Do you get any direction at all from the City, the County or the State of Delaware?
A. No sir.
Q. . [Y]ou don’t direct the activities within the library itself?
A. No sir.
Q. Are you paid for this position?
A. No, sir.
Q. Are the other members of the Board?
A. No.
Tr. 732-33.
. The majority opinion notes that “[tjhe city’s interest lies in the government function of edu-eating the public by providing ... a public library.” Maj. Op. at 745 (emphasis added).
. I am perplexed by the conflicting views of “sovereign function” which appear in one guise in Hollenbaugh and in another one here. In Hollenbaugh, Judge Aldisert wrote and Judge Gibbons joined in the unanimous opinion which, as noted, stated that library operation was not a sovereign function. Here Judge Al-disert has joined with Judge Gibbons and the majority in representing the contrary. In all fairness it might be said that in this respect Hollenbaugh may no longer be the prevailing view in this Circuit judging from the fact that Chalfant represents an en banc majority whereas Hollenbaugh was only a panel opinion. My confusion in this respect stems from the fact that the majority opinion has not addressed this point directly nor has the majority even adverted to it despite its rather extensive analysis of Hollenbaugh.
. Accord, Schlein v. Milford Hospital, Inc., 561 F.2d 427, 428-29 (2d Cir. 1977) (per cu-riam).
. Parks v. “Mr. Ford,” 556 F.2d 132, 137 (3d Cir. 1977) (en banc) (plurality opinion); id. at 162 (Hunter, J., concurring); accord, Schlein v. Milford Hosp., supra, 561 F.2d at 428 n.5.
. Jackson v. Metropolitan Edison Co., supra, 419 U.S. at 357, 95 S.Ct. 449; Moose Lodge No. 107 v. Irvis, supra, 407 U.S. at 175, 92 S.Ct. 1965.
It is true, as the majority opinion notes, that both Burton and the instant case involved situations in which the state was the lessor and the private entity was the lessee. But while the two situations may be somewhat similar in form, they are radically different in substance. Burton involved an ordinary commercial lease from which the state hoped to earn a profit. In this case, the Wilmington Institute owned the library property prior to 1923. In that year, for reasons which do not appear in the record, the Institute conveyed the library’s property to the City and the City granted a perpetual rent-free lease of the building back to the Institute. 417 F.Supp. at 164. Thus, while the form of a lessor-lessee relationship exists in both Burton and the present case, the substance of a symbiotic relationship is not present here.
. Burton v. Wilmington Parking Authority, supra, 365 U.S. at 725, 81 S.Ct. 856.
. See also Wilmington Parking Authority v. Ranken, 34 Del.Ch. 439, 445-46, 105 A.2d 614, 618 (Sup.Ct. 1954).
. The majority opinion asserts that the district court improperly applied a “nexus” test to determine the primary question of “whether the employer was governmental or private.” Maj. Op. at 746. In making that assertion the majority claims that the district court “looked at the status of the agent [the Board of Managers] to determine the nature of the principal [the Institute].” Id I read Judge Latchum’s opinion differently. I believe Judge Latchum has faithfully followed and applied the state action teachings of the Supreme Court and of our Circuit, e. g., Magill v. Avonworth Baseball Conference, 516 F.2d 1328, 1334 (3d Cir. 1975). The district court, having made detailed findings and having engaged in an extensive legal analysis that no indicia of governmental involvement in the Institute existed other than funding, stated: “In conclusion, on the present record, the Court finds that no symbiotic quid pro quo relationship existed between the Institute, the City of Wilmington or New Castle County at the time plaintiff was discharged. See Broderick v. The Associated Hospital Service of Philadelphia, supra, [536 F.2d 1,] at 6-8 [3d Cir.].” 417 F.Supp. at 166.
In addition, the district court dealt with Jackson v. Metropolitan Edison Co., supra, stating: “Jackson demands the existence of a nexus between governmental involvement and the challenged action of the non-governmental entity, and it is clear here that at least until 1974, neither the City of Wilmington nor New Castle County probed into the Institute’s budgetary matters or fiscal administration, or meddled in its overall management, let alone into its employment practices.” Id. at 165-66 (footnote omitted).
Having properly analyzed the record in terms of both a Burton “symbiosis” and a Jackson “nexus”, the district court’s approach cannot be faulted.
. Hollenbaugh v. Carnegie Free Library, 545 F.2d 382 (3d Cir. 1976).
. The opinion in Hollenbaugh does not disclose exactly what percentage of the library’s board of trustees was appointed by governmental bodies. The opinion reveals that 11 of the trustees were government officials, but it does not indicate the total number of trustees in office at that time. A total of 24 trustees was authorized, but there were at least four vacancies. See Braden v. University of Pittsburgh, supra, 552 F.2d at 974 n.15 (Garth, J., concurring).
. See 417 F.Supp. at 165-66.
. The majority opinion states:
As in Hollenbaugh, contracts designate the employer as an agent of the government in furnishing public library services to the public. The only relevant factual distinction between this case and Hollenbaugh is the absence of a specific tax earmarked for support of the library. But in place of a specific tax, the state legislature has by legislation decided the level of appropriations from general tax revenues for library services. Thus, both here and in Hollenbaugh, state law defines the level of library services. Moreover, the record here discloses much greater public involvement. Even the structure, organization and management of the Board of Managers is mandated by state legislation. 19 Del.Laws Chaps. 734 and 983. . . .
Maj. Op. at 745. Neither the record nor the Delaware laws to which the majority makes reference supports these facts or permits the conclusion that the Wilmington Institute is an agent of the state. Indeed, the majority’s analysis ignores the express statement found in Hollenbaugh that the Carnegie Library had accepted designation by the school district and the city to act as an agent for those entities. 545 F.2d at 384. As noted in the text above, no such representation of agency is present or was found by the district court here. Nor do post-hoc funding agreements which have not been shown to implicate the period of Chalfant’s discharge establish such an agency relationship. Hence, the majority’s attempt to equate this case (where no governmental authority had authorized the imposition of a library tax, or designated the library as agent), with Hol-lenbaugh (where the court there found that a specific library tax had been imposed, and an *759express agency created), is neither fortunate nor helpful.
. The majority states that a self-perpetuating method of selecting Board members may not be “any less governmental” than a system of electing Board officials by public franchise. Maj. Op. at 746. While I might agree with this proposition if the self-perpetuating board were composed of public officers rather than private individuals, I cannot agree with the scope of this statement in the context of this case. Here it was not the self-perpetuating feature of the Board which the district court found significant. Rather the district court emphasized in its findings that the Board consisted of “private citizens”, 417 F.Supp. at 165 (emphasis added), who were not selected by virtue of holding title to public office. Moreover the record in this case, and in particular the district court’s findings, will not sustain any implication that the mere service of a self-perpetuating private Board converts private action into state action.
. Hollenbaugh, supra, 545 F.2d at 385; accord, Bethel v. Jendoco Construction Corp., 570 F.2d 1168, at 1176-1177 (3d Cir. 1978); cf. United States v. Orleans, 425 U.S. 807, 815, 96 S.Ct. 1971, 48 L.Ed.2d 390 (1976) (in determining whether a community action agency is an agency of the United States as opposed to a governmental contractor for purposes of recovery under the Federal Tort Claims Act, “the question here is not whether the community action agency receives federal money and must comply with federal standards and regulations, but whether its day-to-day operations are supervised by the Federal Government”).
. Judge Gibbons in his legal analysis for the majority cites Kerr v. Enoch Pratt Free Library, 149 F.2d 212, 219 (4th Cir. 1945) (Pratt). He quotes from that decision to justify the conclusion which the majority here has reached. Maj. Op. at 746.
I observe however that if a record similar to Pratt’s had been developed in the context of this case, Judge Latchum’s result undoubtedly would have comported with the result reached in Pratt, and there would have been no occasion for this dissent.
The record in Pratt established that the city’s and county’s influence, control, and operation were all-pervasive, 149 F.2d at 215-17, and thus state action was properly found to exist so as to prevent the continuation of invidious racial discrimination, id. at 218. Here however the district court has found no such city, county or state influence, control, operation or nexus in the context of an employment dispute. 417 F.Supp. at 163-65. It is because of this essential difference that the majority’s reference to Pratt does not bolster its conclusion.
. Hollenbaugh is distinguishable for another, equally significant, reason. Since the Hollen-baugh court was reviewing an order disposing of a motion for summary judgment, it was required to give to the employees — the parties opposing the motion — “the benefit of all reasonable doubts and inferences.” 10 C. Wright & A. Miller, Federal Practice and Procedure § 2725, at 510 (1971). Viewed in that light, the evidence of the extensive governmental funding which occurred in Hollenbaugh was probably enough by itself to defeat the library’s summary judgment motion. Where the government has provided nearly 100 per cent of a private institution’s financial requirements, certainly it is reasonable to assume at the summary judgment stage that the complaint should not be dismissed because of an absence of state action. See Braden v. Univ. of Pittsburgh, supra, 552 F.2d at 974 (Garth, J., concurring). But once past the summary judgment stage, it is the findings respecting state action which control and any presumptions dependent upon financing are no longer relevant.
Similarly, precisely because Braden presented an appeal from a denial of a summary judgment motion made by the defendant, its holding does not have the breadth attributed to it by the majority opinion. Maj. Op. at 744. Braden holds no more than at that procedural stage, facts which demonstrate (1) massive government financing of the defendant, 552 F.2d at 960; (2) a legislative designation of the defendant as an “instrumentality” of the state, id. at 959; and (3) a possible surrender by defendant of “much of its fiscal autonomy”, id. at 960, will suffice to defeat defendant’s motion *760for summary judgment. Such a showing however can be refuted by the defendant at trial, as happened sub judice. See 552 F.2d at 956 (“[b]ecause the burdens that may radiate from a decision upholding [a finding of state action] may be potentially quite far-reaching, this issue should be dealt with in a full and thoughtful fashion”).
. Moor v. County of Alameda, 411 U.S. 693, 716, 93 S.Ct. 1785, 1799, 36 L.Ed.2d 596 (1973).
. Id.
. In light of my position on this issue, I do not consider the pendent state claims in the context of Aldinger v. Howard, 427 U.S. 1, 96 S.Ct. 2413, 49 L.Ed.2d 276 (1976).