Congress in enacting the National Labor Relations Act gave courts the power to review orders of the National Labor Relations Board (hereafter “the Board”). Findings with respect to questions of fact when supported by substantial evidence shall be left to the discretion of the Board. The Board, of course, is in a position neither to impose certain rules upon parties to a labor dispute while the proceedings are under way nor to abandon its impartiality in aid of one party or the other. In the instant case it would not be difficult for the Court to find in favor of the Union, nevertheless, based on the law we are compelled to disagree with the Board’s decision to so find and we remand the dispute for further consideration.
The present controversy began when the United Steelworkers of America, AFL-CIO (hereafter “the Union”) was certified on October 3, 1966, as the collective bargaining representative of the production and maintenance workers employed by Florida Machine and Foundry Company and Fleco Corporation (hereafter “the Company” or “the Employer”). The Union in due course submitted a contract proposal to the Company which then responded with a counter-proposal; these were analyzed and discussed at a number of negotiating sessions during which the position of each party emerged and began to take shape. After about five months of negotiation, however, the Union apparently *1007despaired of succeeding through this channel and a strike was called for February 28, 1967. Meetings continued during the strike and the parties came closer to a contract, but meanwhile, the Company began to replace the striking workers and, before too long, the strike faltered. Less than five months after it had begun, the strike ended and the Union requested permission for its members to return to work. Thereafter, the Union filed charges with the Board alleging that the Company had violated the National Labor Relations. Act by refusing to bargain with the Union in good faith in its effort to negotiate a contract acceptable to both parties. The Union also charged that the Employer wrongfully refused to reinstate employees who had struck in protest against the Employer’s refusal to bargain in good faith.
A Trial Examiner presided over a hearing at which the parties introduced evidence and interrogated witnesses on the issues raised in the Union’s charges. With a few exceptions not immediately relevant, the Board ratified the Trial Examiner’s findings and accepted his conclusion that the Company had failed to bargain in good faith. An order was issued instructing the Company to cease and desist from this behavior and to offer certain of the strikers immediate and full reinstatement to their former jobs. This was the background of the ease as it came to us upon the Board’s petition for enforcement of its order and the Union’s appeal from a limited portion of the Board’s order concerning the amount of back pay owed to reinstated employees. In view of our disposition of the former issue, we have not considered the latter since the question of back pay arises only after a finding that the Company violated its obligation to bargain in good faith. The errors into which the Board fell while making its decision cast sufficient doubt on that finding to require a remand for reconsideration in line with the observations which follow.
In drafting the order now before us, the Board relied primarily on the Trial Examiner’s finding that the Employer had violated sections 8(a) (1) and 8(a) (5) of the National Labor Relations Act.1 The Examiner summarized his conclusion as follows:
I find upon the entire record that although the Company met with the Union and exchanged contract proposals, it did not bargain in good faith, as required by the Act, that is “with an open and fair mind, and a sincere purpose to find a basis of agreement touching wages and hours and conditions of labor.”
(App. 40.) Since the Employer met with the Union and actively worked on contract proposals, it is necessary to take a very close look at the behavior of the Employer during negotiations and at its proposals before reaching the conclusion that the Employer had no desire to reach an agreement. Accordingly, the data which led the Examiner to a finding of bad faith is the quarry of the present search.
There is evidence in the record from which to conclude that the Employer was not acting in good faith. It was apparent, for example, that the Employer from the beginning had not wanted a union and had done its best to persuade the employees to vote against union representation. Even after the Union was certified as the collective bargaining agent, a residue of this antipathy remained ; the Board agreed with the Ex*1008aminer that there was evidence of this hostility toward the Union in the harshness of the contract terms the Employer proposed. Further support for the bad faith thesis was gained from the Employer’s long delay in supplying data on which certain wage offers purportedly were based and from a refusal to provide certain other information in the form requested by the Union. There was also evidence of an informal (and highly inappropriate) offer, on the eve of the strike, of a ten cent per hour raise for those workers who stayed on the job. Finally, there was at least some evidence from which to infer, as the Board did, that the Employer dragged its feet when urged by the Union to schedule frequent negotiating sessions.
The foregoing should make it clear that the record reflects no glory on this Employer; we are in no way tempted to commend the Company for its part in this proceeding. In the arguments to this court it is suggested that the Employer was motivated by a concern for the best interests of its employees and thought these best could be served without the intercession of a union. We are not so naive as to accept such an obvious and juvenile misrepresentation.
At the same time it is well to point out that the negotiation of labor contracts is not a gentle art. Both labor and management have a great deal at stake when sitting down at a bargaining table and this is reflected in the language used and in the proposals advanced for consideration. This court and the Board each must remain aware of how difficult it is to capture the tone of such proceedings from an outside vantage point. What may appear to be an unreasonable, obdurate demand may be no more than the skillful practice of the negotiator’s art, designed to wring concessions from the opposite side. Congress has excluded the Board from interfering in this process no matter how strongly it feels about the merits of the proposals under discussion.
Notwithstanding these reservations, it might appear that this court reaches the end of its tether in this area when it finds, as we have, that there is evidence in the record to support the Board’s order. The cognoscenti might anticipate first a dutiful nod in the direction of “substantial evidence”—yet another attempt to trap that elusive butterfly with a net of mathematical precision. Ordinarily that would be our course, for we are neither inclined nor empowered to upset administrative findings which might be found to be based on substantial evidence in the record as a whole. N.L.R.B. v. Herman Sausage Co., 275 F.2d 229, 231 (5th Cir. 1960). This is not the ordinary case, however, because we have detected several misconceptions entertained by the Board as to its proper role in this labor dispute. These errors may well have infected the process by which the Board found the Employer guilty of unfair labor practices and of a failure to bargain in good faith; in fairness to the Employer, however reprehensible its behavior, we must remand the proceedings to the Board for reconsideration.
Subtle distinctions, such as between the correct and incorrect role of the Board in labor controversies, are difficult to ferret out of a record of substantial length; and, even if such first step succeeds, it is even more difficult to present a concise, composite picture of what is erroneous in contrast to what would have been correct. Nevertheless, that procedure is a present necessity and one in which we will begin with perhaps the most visible of the distinctions we hope to maintain.
Congress might well have given the Board power to determine the substantive terms of labor contracts, for the Board in the exercise of its expertise would be in a sound position to weigh the competing interests of the parties and to draft terms fair to both. As it happens, however, Congress explicity reserved that power unto the parties themselves. Porter Co. v. N.L.R.B., 397 U.S. 99, 90 S.Ct. 821, 25 L.Ed.2d 146 (1970). *1009The Board is authorized to oversee and enforce the duty of employer and union “to bargain collectively * * * in good faith,” but Congress has made it clear that the obligation of good faith bargaining “does not compel either party to agree to a proposal or require the making of a concession.” 29 U.S.C. § 158(d) (1964). Since the statute involves no compulsion on either party to make a concession or agree to a proposal, then obviously the agency designated to execute the statute is without power to effect a similar compulsion. N.L.R.B. v. American National Ins. Co., 343 U.S. 395, 72 S.Ct. 824, 96 L.Ed. 1027 (1952). Thus it is a fundamental proposition of labor law that the Board’s power in this sphere is rigorously limited; in fact, Congress added section 8(d) to the Act precisely because “criticism of the Board’s application of the [bargaining in good faith] test arose from the belief that [the Board] was forcing employers to yield to union demands if they were to avoid a successful charge of unfair labor practice.” N.L.R.B. v. Insurance Agents’ International, 361 U.S. 477, 486, 80 S.Ct. 419, 425, 4 L.Ed.2d 454 (1960). See also H.R.Rep.No.245, 80th Cong., 1st Sess. 19 (1947).
This leads us to the crux of the problem in the instant case. If we accept the proposition “that [section] 8(d) was an attempt by Congress to prevent the Board from controlling the settling of the terms of collective bargaining agreements” (N.L.R.B. v. Insurance Agents’ International, supra, 361 U.S. at 487, 80 S.Ct. at 426), then we cannot give our approbation to the Board’s finding that the Employer was guilty of bad faith bargaining if that finding is based on the fact that the Board did not approve of certain of the proposals advanced by the Employer in the course of negotiations. It appears, however, that this was the effect of the Board’s adoption of the Examiner’s finding that certain contract terms offered by the Employer were so inherently obnoxious to the best interests of the Union as to constitute evidence of bad faith and an unwillingness to come to agreement. The Examiner found that “[t]he Company proposed and insisted throughout the negotiations that the Union accept contract clauses which drastically curtailed the Union’s representation rights.” (App. 41.) In its decision the Board embraced this ruling without comment, although it did retreat to a limited extent from the implications submerged within it. For instance, the Board noted that it did not adopt the Examiner’s conclusion that the Employer’s refusal to agree to a term providing .for compulsory arbitration and the Employer’s concurrent proposal for a “no-strike” clause was, as a practical matter, an “insist [ence] that the Union virtually surrender its rights to strike without the usual ‘quid pro quo of the compulsory arbitration of grievances.’ ” (App. 57.)
The Board, however, was not always so careful to observe the distinction between, on the one hand, a permissible inference of bad faith drawn from the proposals the Employer advanced and, on the other hand, an impermissible attempt to compel the Employer to accept certain terms or make certain concessions, an attempt in violation of section 8(d) of the Act. In its brief, for example, the Board urges that “[f]rom a substantive standpoint, * * * the Company’s proposal was so disadvantageous to the employees as to call into question the Company’s willingness to sign any agreement.” (Brief for N.L.R.B. at 23.) After enumerating some of the heinous contract terms advanced by the Employer, the Board concludes, in virtually the same words the Examiner used:
Since the Company included in its proposals no improvements in existing conditions of employment which might compensate the employees for this sweeping relinquishment by the Union of their right to representation, its proffered contract was one which the Union could not possibly jústify to the employees and hence frustrated rather *1010than furthered collective bargaining.
(Id. at 24.)
We think that the second statement differs significantly from the first and that the Board there fell on the wrong side of the line dividing “permissible inference” from “impermissible compulsion.” As to the first statement, it may be assumed for immediate purposes that the Board was on safe ground in examining a proposal and finding it “so disadvantageous * * * as to call into question the Company’s willingness to sign any agreement.” (Id. at 23.) We agree, as the Examiner found, that “good faith or its lack is a question of fact as to state of mind, and positions taken at the bargaining table, considered in the context of the whole case, are manifestations of the state of mind with which negotiations are conducted.” (App. 42.) Thus it would be legitimate for the Board to infer some degree of bad faith from the Company’s insistence on a particularly disadvantageous proposal.
As to the second statement quoted above, however, to the effect that one employer concession might “compensate” the employees for a concession they have made through their representatives, the Board is caught up with more than a mere inference as to the Employer’s state of mind. Instead, the Board is dictating substantive negotiating terms in direct violation of section 8(d). When the Board bases a finding of bad faith on the fact that the Employer has not made a certain concession to the Union, then the Board is adding to the obligation of good faith bargaining a requirement that certain concessions be made. We cannot allow the Employer’s obligation to be so altered.
This argument proceeds admittedly on a fine distinction, but the distinction was drawn not by us but by Congress and we have no alternative but to enforce it to the best of our ability. In so doing, we must rely as much on nuances in the Board’s language as we would ordinarily rely on concrete facts; that is yet another reason why this kind of distinction is particularly difficult to articulate. We are not the first to approach this hurdle nor are we the first to point out the “tension between the principle that the parties need not contract on any specific terms and a practical enforcement of the principle that they are bound to deal with each other in a serious attempt to resolve differences and reach a common ground.” N.L.R.B. v. Insurance Agents’ International, 361 U.S. 477, 486, 80 S.Ct. 419, 425, 4 L.Ed.2d 454 (1960).
Having made these observations, we would prefer to be in a position to offer a clear set of guidelines for the Board’s future use, but the nature of our power to review does not allow us that prerogative. We hold merely that in the particular circumstances of this case the Board permitted its view on the merits of the substantive terms at issue between the parties to cloud its judgment of the Employer’s good faith. See Collins & Aikman Corp. v. N.L.R.B., 395 F.2d 277 (4th Cir. 1968). Thus the Board erred by adding to the Employer’s obligation to bargain with the Union in good faith an obligation to concede to the Union’s demands for certain contractual terms. We discern ample support for our holding in the following passage from Porter Co. v. N.L.R.B., 397 U.S. 99, 107-108, 90 S.Ct. 821, 825-826, 25 L.Ed.2d 146 (1970):
It is implicit in the entire structure of the Act that the Board acts to oversee and referee the process of collective bargaining, leaving the results of the contest to the bargaining strengths of the parties. * * * The Board’s * * * powers * * * are limited to carrying out the policies of the Act itself. One of these fundamental policies is freedom of contract. While the parties’ freedom of contract is not absolute under the Act, allowing the Board to compel agreement when the parties themselves are unable to do so would violate the fundamental premise on which the Act is based—private bargaining under governmental supervision of the procedure alone, without *1011any official compulsion over the actual terms of the contract.
(Footnotes omitted.)
A second example of our dissatisfaction with the Board’s order involves the question of wages, one of the major areas of disagreement between the parties and important because in dealing with the problem the Board again misconceived its role. Because the issue of proposed wage increases will recur in virtually every labor dispute, it is particularly necessary that the Examiner and the Board treat the problems arising in this area with consummate skill supported by a complete record and strictly in accord with both statutory requirements and judicial guidelines. In the instant case the Employer offered an across-the-board wage increase of 8 cents per hour and offered additional selective raises for certain specialized jobs. The Union began with a proposal for a 22 cent across-the-board raise but later reduced this figure first to 20 cents and then to 18 cents. In dealing with the parties’ failure to come to agreement on this question, the Board adopted the Examiner’s findings, which were as follows:
As to wages, the Company initially rejected the Union’s wage demands in their entirety on the ground that its wage rates were equal to or better than those in the Jacksonville area. Allegedly based on a new area survey, the Company on January 6, 1967, offered to increase the wages of several skilled job classifications to keep them in line with rates for comparable classifications in the area and, on January 25, it offered to increase shift differentials by 2 cents and give a general wage increase of 8 cents. The Company thereafter refused to supply the Union with area wage survey data. After the strike, on October 2, 1967, it unilaterally put its proposed wage increase into effect allegedly because it was at a “competitive disadvantage in the local labor market” and was not satisfied with “the calibre of applicants.”
(App. 44.) The Examiner concluded that the Employer’s wage offers amounted to “no more than ‘surface bargaining’ and were part of ‘a purposeful strategy to make bargaining futile or fail.’ ” (Id.)
The Examiner also found that the Employer had made “no meaningful concessions on any major issue.” (Id.) If the question of wages was a major issue, and it seems indisputable that it was, then the Examiner must have found that the Employer had made “no meaningful concession” on wages. The Examiner therefore must have believed that a raise of 8 cents an hour was not meaningful and was an example of mere “surface bargaining.” He of course is entitled to make this kind of finding and he is free to draw from it an inference of bad faith, but it is an elementary principle that the Examiner cannot do this without some evidence in the record to support his conclusion. We have sought in vain for such evidence. To the extent that the record sheds any illumination on this question it seems to support the contrary view that the wages already being paid were comparable to those offered by the other employers in the Jacksonville area. The following testimony by one of the Union’s leaders is illustrative of that point:
[Counsel for the Employer] ask[ed] us if we had checked the Jacksonville area rates, and I believe I advised him that was one of our major problems—by “our,” the working people of—the Union, that the Jacksonville rates were substandard.
And, [Counsel] said that the Company had to compete in the Jacksonville labor market, and they were paying a good wage.
And, I asked him how he would compete—how he would compete more poorly if he gave raises, and I don’t recall his answer to that.
(App. 421) (Emphasis added.)
From the absence in the record of any evidence on the exact wages being paid in the Jacksonville area, we are forced *1012to conclude either that the Board assumed that the Employer’s wage offer was not a meaningful concession or that the Board imposes an obligation on the Employer to produce evidence proving that its offer is meaningful and penalizes the Employer if it fails to produce such evidence. We have no way of knowing whether the 8 cent offer was “meaningful” by any objective standard, and it is not our function to speculate. Certainly we cannot draw any conclusion from the fact that the Union reduced its proposal from 22 cents to 18 cents whereas the Employer maintained his 8 cent offer; the Union, as a tactical move, may have begun with an extremely unrealistic figure while the Employer may have decided as a matter of bargaining strategy to begin with its maximum offer. We are in no position to rely on such speculation without evidence to support it, but neither is the Board entitled to speculate as it appears to have done here.
It might be argued that the lack of evidence in the record as to wages in the area of the Company was attributable to the Employer’s failure to supply the Union with a requested wage survey. Although the record is not clear on this point, it appears possible that the wage survey which the Union requested was the survey relied on by the Employer in formulating raises for four specialized job classifications and not a survey which would throw light on the general wage increase with which we are now concerned. The survey covering the specialized jobs was supplied to the Union and is a part of the record, but it is not relevant to the more general problem. The question then becomes whether the Employer had a general obligation to supply evidence to support the reasonableness of its wage offer. We think that this is not the same question as is involved when an Employer has specific wage surveys which it uses while negotiating and yet refuses to provide to the Union. See N.L.R.B. v. Acme Industrial Co., 385 U.S. 432, 87 S.Ct. 565, 17 L.Ed.2d 495 (1967). Our objection is to the proposition, implicit in the Examiner’s findings, that the Employer has a general obligation to prove that its wage proposal constitutes a meaningful concession whereas the Union has no such obligation. We see no justification in the statute for such a disparity in the burden of each party. In the' absence of proof that the Employer has made wage surveys which it readily could make available to the Union, we hold that the Examiner must solicit such information from the parties before he can make any finding on the reasonableness of either party’s wage offer or conclude that either party’s offer is evidence of bad faith.
The Board found two specific violations of the Employer’s collective bargaining obligation, both of which involved information requested by the Union. The Employer was found to have delayed unreasonably in supplying a certain area wage rate survey, which has previously been discussed, and to have refused to supply certain other information accessible only to the Employer. These findings and the conclusion that the strike was not an economic strike but rather was an unfair labor practice strike may or may not stand up after the Board’s reconsideration of the issues already outlined. Since we think a reevaluation of the entire controversy is required on remand, we think it best to postpone discussion of the Board’s two specific findings until such time as it is necessary to take them up. Until the Board has reconsidered the questions presented by this opinion, we are not inclined to comment further on any other aspects of the proceedings.
Remanded for consideration consistent with the views expressed herein.
. The relevant provisions of the Act are as follows:
§ 158. Unfair labor practices.
(a) It shall be an unfair labor practice for an employer—
(1) to interfere with, restrain, or coerce employees in the exercise of the rights guaranteed in [section providing for collective bargaining] ;
* * * * *
(5) to refuse to bargain collectively with the representatives of his employees, subject to the provisions of [section setting out specific rights and obligations],
29 U.S.C. §§ 158(a) (1), (5) (1964).