This is a class action arising under Title VII of the Civil Rights Act of 1964, 42 U.S.C. § 2000e et seq. Plaintiffs are representative members of the class composed of all Negroes employed at defendant’s Roanoke Rapids plant and all Negroes who may hereafter apply for employment at the Roanoke Rapids plant. The multitude of defendants is occasioned by changes in corporate ownership which occurred subsequent to the institution of the action. Since the Roanoke Rapids operation has at all pertinent times been called the Albemarle *137Paper Company, the corporate defendants will hereinafter be referred to as Albemarle.
The district court found that Albe-marle practiced racially disciminatory employment practices prior to July 2, 1965, and that the effect of that discrimination had been perpetuated through a job seniority system. Thus, the district court enjoined Albemarle and the defendant union, Halifax Local 425, United Papermakers and Paper-workers, AFL-CIO, from discriminating against Negro employees, and ordered that the job seniority system be abolished and a plant-wide seniority system be implemented. There has been no appeal from these provisions of the court’s decree.
The district court refused to order the abolition of or changes in the pre-em-ployment testing procedures used by Al-bemarle. Plaintiffs appeal from the district court’s determination. Judge Bore-man concurs with Judge Craven in reversing and remanding to the district court on this issue. Judge Bryan dissents.
The district court also refused to award the plaintiffs back pay. Judge Bryan concurs with Judge Craven in reversing the district court on this issue. Judge Boreman dissents.
The effect of this division in the court is to reverse and remand the district court’s determination as to the testing procedures and the refusal to award back pay.
I
TESTING PROCEDURES
Prior to 1958, no employment personnel tests were given to applicants for employment at Albemarle. In 1956, the personnel manager was requested to design a screening program for selection of employees for certain departments. In this connection, the Revised Beta Examination (Beta) and the Bennett Mechanical Comprehension were selected to test applicants. A brief study was made at that time to determine the usefulness of the Beta test.
About 1963, the then personnel manager discontinued using the Bennett Mechanical Test since it had not been studied. Use of the Wonderlic Test, A and B series, was initiated at that time. Whereas the Beta is a non-verbal test developed to measure the intelligence of illiterate and non-English speaking individuals, the Wonderlic Tests are verbal tests of general mental ability. Use of the Wonderlic Tests was adopted in connection with the Beta because Albemarle felt it essential for new employees to have a certain level of verbal facility because of the increasing technical nature of the operation and the increasing use of printed instructions in the operation of machinery and the like.
The operations of Albemarle are, like other pulp and paper mills, organized on a departmental basis. For purposes of employee classification and promotion, each department is organized into one or more lines of progression. Entrance into each department is effected at the bottom of a line of progression and employees move up, depending on their ability and experience, as vacancies occur. In all, Albemarle has 11 separate departments containing 17 lines of progression. Since 1963, applicants for 8 of these departments and 14 of the lines of progression were required to score successfully on the Beta and Wonderlic pre-employment tests.
After the Supreme Court decision in Griggs v. Duke Power Co., 401 U.S. 424, 91 S.Ct. 849, 28 L.Ed.2d 158 (1971), Albemarle hired an expert in industrial psychology and testing to conduct a validation study of its pre-em-ployment testing procedures. A validation study determines whether test results have a significant relationship to actual performance on the job. The technique adopted for the validation study of Albemarle’s tests was concurrent validation. In concurrent validation, a sample of current employees occupying the jobs, or job groups, in question is selected. These employees are *138then given the tests, and the test scores compared with one or more criteria of the employees’ ability to perform their jobs.
Albemarle’s validation study covered 10 job groups in only 8 of the 14 lines of progression, and 5 of the 8 departments for which the tests are required. The sample of employees was generally selected from the higher level jobs and encompassed approximately 30 percent of the different jobs for which tests are required. The test scores from that sample were compared with two supervisors’ comparative ratings of employees in each job slot. The criteria for the supervisors’ ratings was: “Excluding a man’s attitude, just how well the guy can do the job when he’s feeling right.” (A. 471). No job analysis was done for the jobs in question.
Albemarle’s expert found that one of the tests was validated for 9 of the 10 job groups studied. However, both tests were valid for only one job group.
The effect of the district court’s approval of Abemarle’s testing procedure is to approve a validation study done without job analysis, to allow Albemarle to require tests for 6 lines of progression where there has been no validation study at all, and to allow Albemarle to require a person to pass two tests for entrance into 7 lines of progression when only one of those tests was validated for that line of progression. We think this was error.
Title VII of the Civil Rights Act “proscribes not only overt discrimination but also practices that are fair in form, but discriminatory in operation.” Griggs v. Duke Power Co., 401 U.S. 424, 431, 91 S.Ct. 849, 853, 28 L.Ed.2d 158 (1971). If an employment practice, though facially neutral as the testing procedures here, is shown to have a differential impact on minority employment, it is prohibited unless the employer can prove business necessity. Griggs, at 432, 91 S.Ct. 849. The plaintiffs made a sufficient showing below that Albemarle’s testing procedures have a racial impact.1 It was thus incumbent upon Albemarle to establish business necessity2 by showing that its testing requirements “have a manifest relationship to the employment in question.” Griggs, at 432, 91 S.Ct. at 854. United States v. Jacksonville Terminal Co., 451 F.2d 418, 455-457 (5th Cir. 1971).
While § 703(a)(2), (h) 3 specifically authorizes professionally developed tests not used to discriminate, Griggs makes clear that that section allows only those tests proven to be job related. 401 U.S. at 436, 91 S.Ct. 849. In so holding, the *139Court gave great deference to the interpretation of the Act evinced in a set of guidelines by the enforcing agency, the Equal Employment Opportunity Commission. Griggs, at 433-434, 91 S.Ct. 849. We have also recently noted with approval these guidelines. Robinson v. Lorillard Corp., 444 F.2d 791, 798 n. 7 (4th Cir. 1971).
We think Albemarle has failed in several respects to show that its tests are job related, have a manifest relationship to employment, and have been validated in accordance with EEOC guidelines.
In developing criteria of job performance by which to ascertain the validity of its tests, Albemarle failed to engage in any job analysis. Instead, test results were compared with possibly subjective ratings of supervisors who were given a vague standard by which to judge job performance. Other courts have expressed skepticism about the value of such ill-defined supervisor appraisals. See, e. g., Rowe v. General Motors Corp., 457 F.2d 348, 452 (5th Cir. 1972). Also Cooper and Sobol have stated in their article, Seniority and Testing Under Fair Employment Laws: A General Approach to Objective Criteria of Hiring and Promotion, 82 Harv.L.Rev. 1598, 1662 (1969):
Supervisory ratings, for example, which are possibly the single most common performance measure used in validity studies, are subject to personal prejudice. When test scores are correlated with such ratings, the validation, if it can be called that, is of questionable value and may simply prove that the test has the same bias as the supervisors.
Likewise, in this regard the EEOC guidelines provide:
(3) The work behaviors or other criteria of employee adequacy which the test is intended to predict or identify must be fully described; and, additionally, in the case of rating techniques the appraisal form(s) and instructions to the rater (s) must be included as part of the validation evidence. Such criteria may include measures other than actual work proficiency, such as training time, supervisory ratings, regularity of attendance and tenure. Whatever criteria are used they must represent major or critical work behaviors as revealed by careful job analyses.
(4) In view of the possibility of bias inherent in subjective evaluation, supervisory rating techniques should be carefully developed, and the ratings should be closely examined for evidence of bias.
29 C.F.R. § 1607.5(b) (1970).
We agree that some form of job analysis resulting in specific and objective criteria for supervisory ratings is crucial to a proper concurrent validation study. See, Western Addition Community Organization v. Alioto, 340 F.Supp. 1351, 1354-1355 (N.D.Cal.1972). To require less is to leave the job relatedness requirement largely to the good faith of the employer and his supervisors. The complaining class is entitled to more under the Act.
Even if the validation procedures had been proper, it was error to approve the testing procedures for lines of progression where there had been no validation study. In this case the tests were approved as a requirement for 6 lines, of progression for which the tests had not been validated. While it is true that a test need not always be validated for each job for which it is required, there are narrow limits on when a test may be used without validation. The EEOC guidelines provide:
(2) Where a test is to be used in different units of a multiunit organization and no significant differences exist between units, jobs, and applicant populations, evidence obtained in one unit may suffice for the others. 29 C.F.R. § 1607.4(c) (1970).
*140In this case, a failure to perform job analyses in the lines of progression involved in the validation study and in the other lines of progression for which the tests are required prevents concluding that no significant differences exist in the jobs in question.4
Not only was it error to approve the testing procedures for lines of progression not validated, but it was also error to approve requiring applicants to pass two tests for positions where only one test was validated. Albemarle seeks to justify this under the business necessity rule. Albemarle originally hires all employees into a pool. From this pool the employees move into a line of progression as vacancies occur. Since it is not known into which line of progression an employee will be placed, Albemarle asserts there is nothing wrong with requiring all employees to be qualified for any line.
Albemarle has not shown that hiring all employees into a pool is necessary to the safe and efficient operation of its business, nor has it shown that hiring employees for specific lines of progression is not an acceptable alternative. This they were required to prove to justify their policies under the business necessity test. Robinson v. Lorillard Corp., 444 F.2d 791, 798 (4th Cir. 1971).
Thus we hold that the district court erred in upholding the validity of the pre-employment personnel tests and in refusing to enjoin their use.
II
BACK PAY
Title 42 U.S.C. § 2000e-5(g) provides in relevant part:
The court may enjoin the respondent from engaging in such unlawful employment practice, and order such affirmative action as may be appropriate, which may include reinstatement or hiring of employees, with or without back pay ....
The district court found that Albemarle had practiced discriminatory employment practices and that these practices had been perpetuated through a job seniority system. Consequently, the district court enjoined Albemarle and the defendant union, Halifax Local 425, from continuing the discriminatory practices. Nevertheless, the district court, finding an award of back pay to be within his discretion, refused to award back pay to the complaining class. The reasons given by the district court for refusing such an award were that the claim for back pay was filed nearly five years after the institution of the action, and that there was no evidence of bad faith noncompliance with the Act. *141Neither of these is sufficient to justify the district court’s refusal to award back pay.
In Robinson v. Lorillard Corp., 444 F.2d 791 (4th Cir. 1971), a case which also dealt with a seniority system which perpetuated past discriminatory practices, the defendants made the contentions presented to us here.
The complaint filed by the plaintiffs in the District Court did not specifically request an award of back pay . ' . . At a pretrial hearing concerning the appropriateness of a class action, one of the lawyers for the plaintiffs clearly indicated that the suit was one for injunctive relief rather than recovery of lost wages. It was not until well after trial of the case, though before the judge had entered a decision, that a request was made for additional relief in the form of back pay for the class.
444 F.2d at 802-803. In Robinson, as here:
[Bjeeause the obligation to provide back pay stems from the same source as the obligation to reform the seniority system, any general defenses relevant to the back pay award were equally relevant to the suit for injunc-tive relief .... The defendants have in no way been prejudiced by the belated claim.
444 F.2d at 803.
Other courts have held likewise. The Fifth Circuit held, in United States v. Hayes Int’l Corp., 456 F.2d 112 (1972), that the broad aims of Title VII require that the issue of back pay be fully developed and determined even though it was not raised until the post-trial stage of litigation.
Also, the Seventh Circuit, in a case which was not a class action, remanded the case in order that “ ‘relief should be made available to all who were so damaged whether or not they filed charges and whether or not they joined in the suit.’ ” Sprogis v. United States Air Lines, Inc., 444 F.2d 1194, 1202 (7th Cir.), cert. denied, 404 U.S. 991, 92 S.Ct. 536, 30 L.Ed.2d 543 (1971).
In Robinson the argument was also made that back pay should not be awarded where the defendant had acted in good faith. Put negatively, the argument, urged by Albermarle, is
that back pay should not be awarded in the absence of a specific intent to discriminate. A corollary argument is that the award was improper in light of the unsettled state of the law. The principal answer to both points is that back pay is not a penalty imposed as a sanction for moral turpitude; it is compensation for the tangible economic loss resulting from an unlawful employment practice. Under Title VII the plaintiff class is entitled to compensation for that loss, however benevolent the motives for its imposition.
444 F.2d at 804. See also, Johnson v. Goodyear Tire & Rubber Co., 349 F.Supp. 3, (S.D.Tex., 1972) (also a seniority case where good faith was present).
The only significant difference between Robinson and the present case is that in Robinson the district court had awarded back pay and the district court below refused to award back pay. The question then is whether, in light of the broad aims of Title VII, this court may affirm the opposite result as to back pay on similar factual situations because such award rests in the discretion of the district judge.
“Discretion in a legal sense necessarily is the responsible exercise of official conscience on all the facts of a particular situation in the light of the purpose for which the power exists.” Bowles v. Goebel, 151 F.2d 671, 674 (8th Cir. 1945) (emphasis added). Thus in determining the proper scope of the exercise of discretion, the objective sought to be accomplished by the statute must be given great weight. Hecht Co. v. Bowles, 321 U.S. 321, 331, 64 S.Ct. 587, 88 L.Ed. 754 (1944). Where a district court fails to exercise discretion with an eye to the purposes of the Act, it must *142be reversed. Wirtz v. B. B. Saxon Co., 865 F.2d 457 (5th Cir. 1966); Shultz v. Parke, 413 F.2d 1364 (5th Cir. 1969).
“The clear purpose of Title VII is to bring an end to the proscribed discriminatory practices and to make whole, in a pecuniary fashion, those who have suffered by it.” Bowe v. Colgate-Palmolive Co., 416 F.2d 711, 720 (7th Cir. 1969). In light of the congressional purpose, courts should give “a wide scope to the act in order to remedy, as much as possible, the plight of persons who have suffered from discrimination in employment opportunities.” Rowe v. General Motors Corp., 457 F.2d 348, 354 (5th Cir. 1972).
Also, as we stated in Robinson, “[t]he back pay award is not punitive in nature, but equitable — intended to restore the recipients to their rightful economic status absent the effects of the unlawful discrimination.” 444 F.2d at 802. See also, Johnson v. Georgia Highway Express, Inc., 417 F.2d 1122, 1125 (5th Cir. 1969).
Because of the compensatory nature of a back pay award and the strong congressional policy embodied in Title VII, a district court must exercise its discretion as to back pay in the same manner it must exercise discretion as to attorney fees under Title II of the Civil Rights Act. See Bowe, supra 416 F.2d at 719. Thus, a plaintiff or a complaining class who is successful in obtaining an injunction under Title VII of the Act should ordinarily be awarded back pay unless special circumstances would render such an award unjust.5 Newman v. Piggie Park Enterprises, 390 U.S. 400, 88 S.Ct. 964, 19 L.Ed.2d 1263 (1968). See Bowe, supra 416 F.2d at 719-720.
Because there are no such special circumstances here, on remand the district court should include an award of back pay in its order. It is to be remembered, of course, that a back pay award is limited to damages which are actually suffered. Lea v. Cone Mills Corp., 438 F.2d 86 (4th Cir. 1971).
Reversed.
. It was demonstrated in plaintiff’s exhibits that on the Wonderlic Series 96 percent of whites passed as opposed to 64 percent of blacks. This corresponds with . the national average as determined in a study by the authors of the test. Negro Norms, A Study of 38,452 Job Applicants for Affirmative Action Programs, E. F. Wonderlic & Assoc., Inc. (1970).
. “[T]he applicable test is not merely whether there exists a business purpose for adhering to a challenged practice. The test is whether there exists an overriding legitimate business purpose such that the practice is necessary to the safe and efficient operation of the business. Thus, the business purpose must be sufficiently compelling to override any racial impact; the challenged practice must effectively carry out the business purpose it is alleged to serve; and there must be available no acceptable alternative policies or practices which would better accomplish the business purpose advanced, or accomplish it equally well with a lesser differential impact.”
Robinson v. Lorillard Corp., 444 F.2d 791, 798 (4th Cir. 1971). See also, Note, Employment Testing: The Aftermath of Griggs v. Duke Power Company, 72 Colum.L.Rev. 900, 908 (1972).
. § 703(a) (2), (h) provides in relevant part:
“Notwithstanding any other provision of this subchapter, it shall not be an unlawful employment practice for an employer ... to give and to act upon the results of any professionally developed ability test provided that such test, its administration or action upon the results is not designed, intended or used to discriminate because of race, color, religion, sex or national origin.” 42 U.S.C.A. § 2000e-2(h).
. The appellants also question the validity of the validation study because the employees tested were primarily from higher level, rather than entry level, positions. As pointed out by the EEOC guidelines, 29 C.F.R. § 1607.4(c)(1) (1970), such a validation technique is permissible only where new employees will progress to the higher level “within a reasonable time and in a great majority of cases.” As stated by Cooper & Sobol, Seniority and Testing under Fair Employment Laws: A General Approach to Objective Criteria of Hiring and Promotion, 82 Harv.L.Rev. 1598, 1649 (1969):
In addition, the problems inherent in test use are compounded by the fact that the higher level jobs may not be reached for several years, during which time an employee may improve his capabilities. This possibility seems particularly likely in the case of an individual whose potential has been artificially suppressed through discrimination, in education and otherwise. In order for a promotability interest to justify test use, therefore, it is essential that the test have been properly evaluated and shown to be predictive of performance at some job level to which promotions can be expected in the reasonable future.
In Griggs the Court failed to reach the question of whether a testing procedure which takes into account higher level job requirements is permissible. 401 U.S. at 432, 91 S.Ct. 849. We likewise fail to reach this question because of insufficient evidence below as to the time in which it takes an employee to progress to higher level jobs.
. For examples of such special circumstances said to justify refusal of a back pay award, see Schaeffer v. San Diego Yellow Cabs, Inc., 462 F.2d 1002 (9th Cir., June 20, 1972); Le Blanc v. Southern Bell Tel. & Tel. Co., 333 F. Supp. 602, 610-611 (E.D.La.1971), aff’d, per curiam, 460 F.2d 1228 (5th Cir. 1972).