This appeal presents issues of segregated facilities and programs that were long ago resolved in the courts of this country. The case also raises issues related to job assignment, transfer, promotion, training, recruitment, seniority, and testing; some of the answers to these questions seem clear, but others are still being formulated by legal processes. All of the issues, the settled and the unsettled, are intertwined.
I.
STATEMENT OF THE CASE
On October 5, 1966, the named plaintiffs, Patrick James, Howard Harville and Louis Winston, black employees at Stockham’s Birmingham facilities, filed charges of discrimination with the Equal Employment Opportunity Commission (“EEOC”), against Stockham Valves and Fittings, Inc. (“Stock-ham” or “company”), alleging that the company maintained racially segregated facilities; discriminated against black employees in job assignment, promotion, training, and transfer; and employed discriminatory testing, education, and age requirements. The EEOC found “reasonable cause” to believe that Stockham engaged in discriminatory practices and issued the plaintiffs a “right to sue” notice in February 1970.
The plaintiffs brought this class action suit on March 16, 1970, within the thirty-day statutory period, against Stockham under the Civil Rights Act of 1866, 42 U.S.C. § 1981, and Title VII of the Civil Rights Act of 1964, 42 U.S.C. § 2000e et seq. The plaintiffs filed an amended charge of discrimination with the EEOC on June 8,1970, against the United Steelworkers of America, AFL-CIO (“Steelworkers”) and its Local 3036 (“Local” or “Union”), and later amended its complaint by adding the Steelworkers and the Local as defendants. The union defendants are alleged to have violated Title VII and 29 U.S.C. §§ 151 et seq. (“the duty of fair representation”). The district court referred the case to the EEOC for conciliation until June 1973 when the district court granted the plaintiffs’ motion to set aside the stay order. The court certified the class represented by the named *314plaintiffs under Rule 23(b)(1), F.R.Civ.P., to include all black hourly production and maintenance employees of Stockham who are currently employed and all black persons who have been so employed at Stock-ham from July 2, 1965, to the date of trial. The trial was held from February 4 through February 22, 1974.
The district court rendered final judgment March 19, 1975. Relying heavily on the proposed findings of fact and conclusions of law filed by the defendant Stock-ham,1 the court found that, except for those segregated facilities maintained by Stock-ham and resolved in a conciliation agreement between the EEOC and Stockham two weeks before the trial, Stockham had engaged in no employment discrimination.2 The plaintiffs appeal from the district court’s judgment in favor of the defendants.3
II.
FACTS
A. Introduction
Stockham is engaged in the manufacture of cast iron valves; malleable fittings; bronze, iron and steel valves; and other industrial valves and fittings at its facilities in Birmingham, Alabama. The product diversity and overall capacity of the company have gradually increased since Stockham *315was founded in 1903. By 1973 Stockham’s work force in Birmingham was comprised of more than two thousand employees. Although the district court found that “[h]istorically, approximately two-thirds of Stockham’s employees have been black”, 394 P.Supp. at 443, the record reveals that the two-thirds figure applies to production and maintenance workers during the years from 1966 to 1973. Approximately 56 percent of the entire work force at Stockham’s Birmingham facility was black during that same period, a figure larger than the percentage of blacks in the Birmingham area.
The district court found that Stockham has a multi-plant complex in Birmingham that is, in effect, six plants in one, comprised of a cast iron fittings plant, a malleable iron fittings plant, a bronze valve plant, an iron valve plant, a steel valve plant, and a butterfly valve plant. That finding is overstated, for at least four of the twenty-two seniority departments at Stockham, valve machining and assembly, electrical, machine shop, and construction, extend over all or virtually all of the “plants”.
The defendant unions, the Steelworkers and its Local, have been the bargaining unit representatives for the production and maintenance hourly employees at Stockham since 1944. A majority of the local union’s members have been black since World War II, and a majority of the members of the Local’s grievance committee and of its officers have been black since 1967. Plaintiffs James and Winston have been officers of the Local and participated in collective bargaining negotiations. The Steelworkers’ staff representative who has aided the Local in contract negotiations is black.
All the plaintiffs were black hourly employees of Stockham. Patrick James, a high school graduate and a graduate of Booker T. Washington Business College, was hired as a laborer at Stockham in 1950 and twenty-four years later at the time of trial was still working in that capacity. Howard Harville was hired in 1946 and worked as an arbor molder in the grey foundry until 1970 when he retired on a medical disability. Louis Winston was hired as a laborer for the galvanizing department in 1964, was transferred to the electrical department as a laborer in 1965, and in 1971 became one of the first blacks enrolled in the apprenticeship program.
B. Organization
1. Departments
By agreement with the Local, Stockham has maintained a formal departmental seniority system since 1949. There are twenty-two seniority departments. The foundry departments produce the basic materials and molds for Stockham’s products (e. g. grey iron foundry, bronze foundry, and malleable foundry); other departments assemble, finish, and machine products (e. g. tapping room and valve machining and assembly); and another group of departments perform maintenance functions (e. g. electrical shop, machine shop, valve tool room, and construction).
Since 1965 the company has regularly employed approximately two hundred office and clerical personnel. In addition, the work force includes twenty-two non-union, salaried timekeepers. As of June 1973, there were also thirty-two plant guards. The Stockham sales department in Birmingham included twenty-two employees at the time of trial. At that time a total of forty-six salesmen were employed by Stockham throughout the country.
2. Wage Determinants
Within each seniority department bargaining unit jobs are divided into twelve job classes in ascending order of hourly wage from JC 2 to JC 13. These classifications reflect the increasingly complex nature of the jobs and the level of skill necessary to perform them. An employee’s job classification determines his base pay rate. Other factors such as incentive earnings and merit raises also determine actual earnings. For each job classification there are different gradations of pay for non-incentive employees. Under Stockham’s incentive system employees in highly repetitive jobs can add to their base pay if their work *316output reaches a sufficiently high level. A direct incentive worker’s earnings averages approximately twenty-five percent above his base pay rate. Indirect incentive workers provide support services to direct incentive workers and receive incentive pay based on the output of the incentive workers. Non-incentive workers advance from one grade of pay to the next within a job classification if they achieve a predetermined score under a formal merit rating system. Although incentive workers are not eligible for merit pay raises, all employees receive merit ratings from their foremen every six months.
3. Advancement and Transfers
The merit scores received by both incentive and non-incentive employees become part of their personnel records; such ratings constitute one of the factors considered in promotion and training selection.4
Job vacancies have never been posted at Stockham and the company does not have a formal bidding system. In 1965 Stockham instituted a “timely application” procedure that received a formal blessing in the 1970 collective bargaining agreement. An employee may ask his supervisor to prepare an application on his behalf for any job at Stockham, whether or not a vacancy for that job exists at the time of the application. The application is considered “timely” regardless of when the vacancy occurs. In filling vacancies company officials are not restricted to those employees who have filed timely applications. In practice many promotion and training selections are made in favor of employees who have not filed such applications.
Stockham administered the Wonderlic Test (discussed later in this opinion) to job applicants and employees seeking promotions and transfers from August 1965 until April 1971. To be considered for a position an employee was required to attain the Wonderlic score designated for the job. An employee seeking a job in a new department, another department from the one in which he was working, was required to obtain the higher “norm” score on the test; a worker seeking promotion within his own department was eligible for the job if he achieved the lower “minimum” score on the test, provided that he had attained “basic departmental job skills”.
Under the seniority system at Stockham a senior employee is entitled to preference only when two or more competing workers possess the same degree of qualifications. An employee’s foreman decides whether he meets this test and is entitled to promotion or training opportunities. This decision is totally within the discretion of the foreman and is not subject to review.
C. Employment Practices
1. Initial Job Assignments
According to company officials, it has been and still is the practice at Stockham for the initial job assignments of new employees to be made by the supervisors, both superintendents and foremen, of the departments containing the vacancies. The personnel office at Stockham serves as a recruiting agency and interviews and screens job applicants. The supervisor of a department advises the personnel office of any need for additional employees, and he is informed when suitable applications are available for review. In some cases the supervisor will request a particular individual whom he knows has filed an application. In other cases the personnel office will present the supervisor with a group of applications for examination. The supervisor, either the foreman or superintendent depending on- the department, makes the hiring decision, which is totally discretionary *317and without written guidelines. Supervisors usually accept approximately seventy-five percent of the applicants recommended by the personnel department.
2. Seniority System
As stated, a Stockham employee seeking a new job in his existing department or desiring to transfer to a position in another department may file a timely application. A supervisor fills the vacancy within his department from employees who have filed timely applications, other employees, and applicants from outside the company. If two Stockham applicants are about equal in qualifications, the collective bargaining agreement requires that the employee with the most departmental seniority be selected.
If other factors are equal, departmental seniority determines not only promotions but also lay-offs and recalls. A worker who transfers between departments is a new employee for purposes of promotion and regression in the transferee department. Before June 1970 if a worker transferred departments he immediately lost all seniority in his old department. In 1970 this requirement was modified in the collective bargaining agreement. An employee was given eighteen months after transfer to the new department to decide if he wanted to return to his old department. If within that time he decided to return, he would be permitted to reenter his old department within twenty-four months of his transfer with his accumulated seniority. The 1973 collective agreement further modified these seniority provisions. If after eighteen months an employee elected to remain in the transferee department, then he would be allowed to retain his seniority in his old department solely for lay-offs, but only until he had been in the new department as long as he had been in the old. If he was laid off during this period, he would be permitted to return to his old department with his accumulated seniority.
The basic features of Stockham’s seniority system have remained unchanged: (1) an employee who transfers between departments forfeits his accumulated seniority at some point; (2) an employee who transfers between departments is a new employee for all promotion and regression purposes; and (3) a departmental employee has preference over employees from other departments for promotion to all vacancies within his department.
3. Craft Training
Craft positions, defined by the company as jobs with classifications from JC 10 through JC 13, are filled through on-the-job training and the apprenticeship program. More skilled tradesmen are trained through the apprenticeship program at Stockham than exclusively through on-the-job training. The program at Stockham involves four years and nine thousand hours of training in shops and apprentice classes. There are no formal lines of progression for the craft jobs. The craft skilled maintenance positions include millwright, electrician, carpenter, patternmaker, blacksmith, and machinist. The craft skilled production jobs include box floor molder, ductile melter, oven operator, crane operator, heat treater, and service mechanic.
The foreman or superintendent of the department in which a craft vacancy occurs selects a candidate for the apprenticeship program. Although timely applications are filed for openings in the apprenticeship program, the supervisor may select an employee who has not filed an application. A majority of employees selected for the program between 1965 and 1971 had not filed timely applications. Apart from several specific requirements, there are no formal guidelines for the supervisor’s selection; he considers such general factors as “desire” and “aptitude” for the craft position. The foreman or superintendent recommends his candidate for the apprentice program vacancy to the apprenticeship committee. As a practical matter, the committee approves the supervisor’s selection virtually automatically.
The specific requirements for the apprenticeship program have changed over the years. From August 1965 until April 1971, an applicant for the program was required *318to score at least 18 on the Wonderlic Test. Also, beginning in 1953 employees were required to achieve a passing score on the Bennett Mechanical Test. In addition to the testing requirements, in 1970 the company instituted a thirty-year maximum age limit, excluding time spent in military service, and a requirement that the applicant have a high school education or its equivalent for eligibility in the program. Before 1970 a grammar school education was required. Company officials have the discretion to waive either the age or the high school education requirement and have done so for a few individuals.
4. Supervisory Selection
Supervisors of the seniority departments at Stockham are selected either directly from the hourly work force without specific training or after completion of one of two training programs maintained by the company. The personnel development program (“PDP”), established in 1960, is designed to train the company’s own employees. Before 1969 the program was informal and unstructured; only three classes were held between 1960 and 1969. There is an allocation of positions in the program among the departments at Stockham and final selection of participants is made by superintendents and foremen. There are no formal, written guidelines for the selection of employee participants in the PDP. The management training program (“MTP”) and its predecessor, the organizational apprentice program (“OAP”), were designed to train individuals with technical skills who could eventually assume positions in upper level management. The OAP was instituted in 1950 and replaced by the MTP in 1969. Under both programs Stockham recruited participants only from nearby predominantly white universities such as Auburn University, the University of Alabama, the University of Tennessee, Georgia Institute of Technology, and Samford University.
As for employees selected directly from the hourly work force for supervisory positions, specific training is not a prerequisite. The superintendent of a division that has a supervisory vacancy may select candidates for the position after consultation with the company’s production manager. An employee may make a timely application for a supervisory position, but the overwhelming majority of the employees selected to be supervisors have never filed such an application. A majority of Stockham’s foremen were promoted from the ranks of hourly workers.
A company official testified that the principal criterion for selecting a foreman is — “who is the best man for the job at the particular time?”. Other considerations are the candidate’s desire to be a foreman, work record, knowledge of the job, training, physical fitness, and common sense. Whether the candidate has a high school education is also considered although employees without high school educations have been selected as foremen and superintendents. There are no specific written standards for the selection of supervisors.
5. Testing Program
In recent years at Stockham testing has come to play an increasingly large part in the selection of individuals for initial employment, training programs, promotion, and transfers. A large-scale testing program was initiated at Stockham in August 1965. Before then only the Bennett Mechanical Comprehension Test for screening apprentice program candidates was in use. In 1965 the company instituted the Wonderlic Personnel Test to screen applicants for initial hire, promotion, and transfer. In October 1966 Stockham formally adopted dual scoring standards for transfers to the three groups of job classes. As discussed above, the minimum score for intradepartmental job candidates was set at a lower level for each group of job classes than was the score for transfers between departments. The candidates for interdepartmental transfers were required to obtain the higher “norm” score while the intradepart*319mental job candidates were eligible for consideration if they achieved the “minimum” score, provided that they had developed “basic departmental job skills”.5 The “minimum” score was established by a committee composed of Stockham management personnel and the company attorney. No one in the group and no company official associated with the administration of the testing had training in the testing field. The Wonderlic Test was never validated at Stockham. In 1971, apparently as a result of the Supreme Court’s decision in Griggs v. Duke Power Co., 1971, 401 U.S. 424, 91 S.Ct. 849, 28 L.Ed.2d 158, Stockham terminated its testing program.
In early 1973, however, the company hired Victor Tabaka, a management consultant, to develop new testing procedures for implementation at Stockham. In April 1973 Tabaka submitted his testing proposal to the company. According to Tabaka he validated the use of tests for specific jobs at Stockham by means of a “concurrent criterion” study. The district court found that the Tabaka tests were put into formal use at Stockham on July 17, 1973. But the company official responsible for employee testing testified that the Tabaka test had not been used in any employee selection decision at Stockham by the time of trial. He stated that the test was administered to employees only for the purpose of accumulating data. The Tabaka test was not proposed for use with supervisory, clerical, or other salaried jobs. In addition, Tabaka concluded that the test was not applicable to eighty-six additional job titles because the associated job aptitudes were not measurable by pen and pencil tests.6
Job Class “Minimum” “Norm”
1-5 0 5
6-8 8 15
9-13 15 18
HL
DISCRIMINATION, ADVERSE IMPACT, AND THE BUSINESS NECESSITY DOCTRINE
The plaintiffs-appellants raise a broad spectrum of issues in this appeal, challenging every aspect of the district court’s finding that Stockham and the unions were not guilty of discriminatory employment practices.
A. Segregated Facilities
Segregation of many of Stockham’s facilities and programs continued into the 1970’s and, in some cases, until a few weeks before trial. In 1965 the system of total segregation extended to the entrance gates of the plant, employee identification numbers, pay windows, toilet facilities, the cafeteria, drinking fountains, the locker rooms, and bathhouses. In addition, the company-sponsored Young Men’s Christian Association (“YMCA”) had two boards and separate Bible classes, one board and one Bible class composed of white members and the other board and Bible class of black members. In addition, there were racially separate company baseball and bowling teams.
Until 1969 all black hourly employees at Stockham had badge identification numbers ranging from 300 to 2999, while all white employees had badge numbers above 2999. Employees with numbers up to 2999, that is, the blacks, used one set of pay windows for cash payments; whereas, white employees with numbers greater than 2999 used another set of windows. In 1969 the company began to assign badge identification numbers by departments and in 1972 abandoned cash payments and the use of pay windows.
The partitions in the cafeteria,7 toilets, and bathhouse,8 the racial assignment of *320lockers, and the racially separate YMCA boards were the subjects of a conciliation agreement between Stockham and the EEOC entered into January 21, 1974. In addition, the plaintiffs adduced testimony at trial on the existence of racially separate
toilet facilities for the seven female employees in the dispensary at Stockham.9
The district court denied an injunction against the continued segregation of facilities at Stockham because he concluded that the issue was “effectively resolved” by the *321conciliation agreement. 394 F.Supp. at 499. In addition, the court found that as to the two women’s restrooms in the dispensary, the plaintiffs failed to establish racially discriminatory practices. 394 F.Supp. at 481.
The plaintiffs and the amicus EEOC contend that the district court erred in its finding that the conciliation agreement resolved all issues of segregated facilities at Stockham. The evidence of the two racially separate women’s restrooms in the dispensary at Stockham supports this contention. That evidence and the showing of the company’s intransigent resistance to desegregation of plant facilities and programs convinces us that the district court over-relied on the conciliation agreement. The plaintiffs and the EEOC request that this Court grant broad injunctive relief against the continued segregation of facilities at Stockham. We reserve that issue for our discussion in subsection “V.A.” on the injunctive relief appropriate for this case.
B. Job Allocations
The plaintiffs-appellants and the amicus EEOC contend that the district court committed plain error in its finding that “Stockham has at no time made initial job assignments (either to departments or to specific jobs) on the basis of an employee’s race.” 394 F.Supp. at 455. We agree.
The evidence in the record of pre-Act assignment of employees to departments and jobs by race at Stockham is overwhelming. The plaintiffs adduced testimony from a variety of company officials that the racial allocation of jobs was the “general rule”. E. Reeves Sims, the employee relations manager, testified:
Q [Plaintiffs’ attorney] Mr. Sims, do you know of any job prior to 1965 which was manned by both black and white employees?
A [Sims] I can’t remember one.
Q Mr. Sims, I’m referring to your deposition which was taken on the 6th day of November, 1973, to Page 146 and 147 starting on line 17.
A May I see it?
Q I was going to say—
MR. NEWTON: Right there.
Q Now, if I may, I will read the question. The question was, “Was there a time, Mr. Sims, when blacks were initially assigned to some jobs and whites were initially assigned to other jobs as a general rule? Answer: Was there a time? Question: Yes. Answer: Yes”.
A As a general rule, yes, as a general rule. But you asked me another question in a different context, Mr. Goldstein. Q And if we can continue then on 147, “And did this practice continue until 1965? Answer: Yes, sir”.
Now, Mr. Sims, you say as a general rule that was true?
A (Nodding head affirmatively).
Q Can you think of any exceptions to that general rule?
A Not offhand right now, no sir.
Q Now, Mr. Sims, this general rule which we have discussed about there being black jobs and white jobs at the company, is that written down anywhere? A No sir.
Q How was it enforced, or how was it put into practice?
A It was in practice when I came to Stockham, and—
Q Would you just say it was a custom? A Yes, sir, custom.
Harry M. Burns, vice-president for corporate products, confirmed that there were no jobs at Stockham before 1965 in which both black and white employees were working. Norman E. Carlisle, superintendent of the tapping room, and an employee at Stock-ham since 1942, reiterated this fact. Terrell G. Burt, manager of technical services, admitted that prior to 1965 only the best qualified whites were considered for clerical positions at Stockham.10
*322Finally, in addition to the uncontradicted testimony of company officials, the plaintiffs introduced evidence that in June 1965 not one of the several hundred hourly jobs at Stockham was filled by both a black and a white.11 The plaintiffs augmented the evidence of racial staffing by jobs with statistics showing that in 1965 jobs for whole seniority departments were allocated on the basis of race. For example, eight of twenty departments12 were 100 percent black in 1965 and two others were more than 95 percent black while two departments were 100 percent white and two additional ones were more than 85 percent white.13 Further, in those departments staffed by black and white employees blacks were concentrated in jobs with the lowest job classifications. While no blacks worked in jobs classified in job class seven (“JC 7”) or above, 95 percent of the white workers were employed in jobs classed above JC 6.14 The employee relations manager, E. Reeves Sims, also testified that blacks were not hired for clerical positions until 1965.15 In addition, there were no black timekeepers or guards at Stockham before 1965. The salesforce was all white at least until the time of trial.
In short, the record demonstrates that the district court plainly erred in concluding that at no time were initial job assignments made on the basis of race at Stockham. The “custom” of job assignments by race at least until 1965 was established without contrary evidence. The testimony of Stock-ham officials that segregation within and between departments was the “rule” plus the plaintiffs’ statistics on racial staffing discussed above establish conclusively that Stockham engaged in racially motivated job assignments before 1965.
The plaintiffs carry their assertion of discrimination in job assignments into the post-Act period. They contend that all relevant evidence compels the conclusion that discrimination in job assignments continued at least until the time of trial. In support of their contention that racial allocation of employees by departments and by jobs within departments is the usual rule at Stockham the plaintiffs emphasize the patterns revealed by statistical evidence.
*323Chart A bears out this contention. Chart A presents data on the percentage of black employees working in each seniority department in 1973, as compared with the percentage of those employees working in each department in 1965.
CHART A
DEPARTMENTAL EMPLOYEES BY RACE
Seniority Denartments
1965 B W
%B 1965
1973 B_ W_
%B 1973
Galvanizing 9 0 100% 15 0 100%
Coreroom & Yard 24 0 100% 76 1 99%
Grey Iron Foundry 92 0 100% 292 16 95%
Pinal Inspection 16 0 100% 52 4 93%
Malleable 88 4 96% 259 19 93%
Brass Foundry 30 1 97% 59
Shipping 22 0 100% 56
Foundry Inspection 25 0 100% 56 9
Dispatching 4 0 100% 27 7 79%
Brass Core Room 11 0 100% 11 3 79%
Tapping Room 53 16 77% 151 45 77%
Valve Finishing Insp. 8 4 67% 20 18 53%
Construction 5 6 45% 15 18 45%
Valve Machining & Assembly 76 36 68% 70 171 29%
Foundry Repairs 4 10 29% 12 55 18%
Machine Shop 3 9 25% 8 50 14%
Electrical 1 7 12% 2 19 10%
Pattern Shop 1 7 12% 3 37
Valve Tool Room 0 5 0% 1 17 6%
Tapping Tool Room 0 11 0% 2 30
Chart A reveals a slight reduction in racial staffing by departments between 1965 and 1973. Nevertheless, in 1973 eleven of the twenty departments16 were predominantly black, while seven of the departments were predominantly white. Only two departments contained black and white employees in approximately equal numbers. As of September 1973 nine hundred and three or 72 percent of all blacks in the hourly work force worked in the departments of galvanizing, coreroom and yard, grey iron foundry, final inspection, malleable, brass foundry, shipping, foundry inspection, dispatching, and brass coreroom. Only 75 whites or 13 percent of all whites worked in those departments. In September 1973 thirty-six percent or 208 of all whites in the hourly work force and two percent or 28 of all blacks worked in the seniority departments of the tapping tool room, valve tool room, pattern shop, electrical shop, machine shop, and foundry repairs. Even more significantly, of the 162 employees hired since 1965 to work in predominantly white departments, and working as *324of September 1973,147 or 90.7 percent were white; whereas, of the 695 hired since 1965 to work in predominantly black departments, 624 or 89.8 percent were black.17
According to the plaintiffs, individual jobs at Stockham have also continued to be assigned largely on the basis of race. Chart B offers statistical support for this contention.
CHART B
RACIAL STAFFING OF JOBS AT STOCKHAM *
*325The data reveal that while no jobs were staffed by both blacks and whites in any department in 1965, that pattern had improved only slightly by 1973 when only 16 percent of all production and maintenance jobs were integrated. In addition, much of the improvement came between November 1970 and June 1973, after the plaintiffs had brought this action in the district court and long after the initial EEOC charge was filed.18
The plaintiffs also make the point that post-Act racial allocation of job opportunities extended to clerical, timekeeper, and guard positions. Chart C gives the number of black and white clerical employees for selected years between 1966 and 1973.
CHART C
CLERICAL EMPLOYEES BY RACE
1966 1968 1971 1973
White 193 200 184 189
Black 5 6 14 18
% black 2.5 2.9 7.1 8.7
Thus, in 1973 only 8.7 percent of clerical workers were black. In addition, although the company employs 22 timekeepers, only two blacks have ever been chosen for that job. About half of all timekeepers are selected from the hourly production and maintenance work force, which is 66 percent black. As of June 1973 there were 25 white plant guards and seven black guards. All three of the sergeants were white. Finally, in 1973 the company’s Birmingham sales department had twenty-two employees; all were white.
The plaintiffs point out that in production and maintenance departments staffed with white and black employees, the blacks were principally concentrated in jobs with the lowest job classifications whereas whites worked largely in jobs with the highest job classifications. Chart D lists the job classes of white and black employees, working at Stockham in September 1973,19 for four different time periods between June 1965 and June 1973.
CHART D
JOB CLASSES OF WHITE AND BLACK INCENTIVE AND NON-INCENTIVE WORKERS
INCENTIVE WORKERS
Job Class
June 1965 .B. M.
June 1968 .B. W.
Nov. 1970 M. W_
June 1973 JL W.
B9 0 5 0 1 0 11 2 22
B8 0 47 2 53 4 60 18 111
B7 0 4 18 \ 23 1 70 9
*326Job Class
June 1965 B_ W
June 1968 B_ W
Nov. 1970 B_ W
June 1973 B_ WL
B6 25 0 34 0 47 0 64 0
B5 134 0 214 1 235 1 279 11
B4 21 1 72 1 82 0 109 5
B3 89 0 80 0 94 0 102 4
B2 _58 _0 _21 _1 _32 157 _15
TOTALS 327 57 441 58 517 74 801 177
NON-INCENTIVE WORKERS
Job Class
June 1965 B. W_
June 1968 B_ W.
Nov. 1970 B W
June 1973 W_
13 0 46 0 55 4 77 1 144
12 0 9 0 7 0 9 3 34
11 0 10 0 9 0 14 0 22
10 0 5 0 14 0 14 2 30
9 0 24 2 34 3 42 9 52
8 0 2 1 4 3 6 7 7
7 0 2 1 5 2 11 24 24
6 2 9 16 11 19 8 27 25
5 103 3 108 1 129 5 143 9
4 8 0 11 0 15 1 40 1
3 31 0 43 1 45 0 56 5
2 34 0 _59 3 _61 _2 190 _14
TOTALS 178 110 239 144 281 189 502 367
Because base pay rates at Stockham are determined by job classifications, the data revealed in Chart D support the conclusion that lower-paying jobs were allocated to blacks at Stockham, at least until 1973. The plaintiffs emphasize that the job class statistics reveal that of 366 white non-incentive workers, 274, or 75 percent, were in JC 9 or above while only 11, or 3 percent, of the 371 black non-incentive workers were in the higher job classes. Of the 178 white incentive workers, 135, or 76 percent, were in the two highest incentive job classes; whereas, only 20, or 2 percent, of the 872 black incentive workers were in JC 8 and 9. Further, the plaintiffs note that the average job class for blacks was substantially lower than the average job class for whites between 1965 and 1973.20
*327Further, the plaintiffs presented evidence of the disparities in black and white employee wages and gross earnings at Stock-ham. The job class differences between black and white workers also reveal the disparities in base pay for blacks and whites; the pay rate for each job is determined by the job class in which the job is located. Non-incentive workers are paid at an established hourly rate. As of June 1973 the pay range for JC 2 was $2.85 to $3.30 per hour while the range for JC 13 was $3.66 to $4.47 per hour. An incentive worker is paid at a guaranteed base pay rate somewhat below that paid a non-incentive worker in the same job class. For example, the incentive rate for JC 2 is $2.85 per hour while it is $3.29 per hour for JC 9. An indirect incentive worker is guaranteed a slightly higher base pay rate and receives on the average less incentive pay than an employee on the direct incentive system. A direct incentive worker’s pay averages approximately 25 percent above his base rate because of his incentive earnings. The exact amount of incentive pay varies with the productivity of the employee.
The plaintiffs showed that the average hourly earnings rate of black employees, including base, incentive, and overtime pay, as of September 1973 was $3.83, or $0.37 less per hour than the average earnings of white employees. Similarly, white employees averaged approximately 12.8 percent more in yearly gross earnings than black employees during the period from January 1, 1973, through September 1, 1973. These pay disparities existed in 1973 even though blacks had greater seniority on the average than whites. The average hiring year for black employees in September 1973 was late 1963; whereas, white employees had an average hiring year of mid-1965. These statistics on pay disparities demonstrate that blacks have been assigned to jobs with lower economic returns than have white employees.
Finally, the plaintiffs assert that blacks were assigned the least desirable jobs at Stockham both in terms of working conditions and the pressures associated with the work. Otto Carter, a white company superintendent, admitted that the hottest, dirtiest, and dustiest parts of the operation at Stockham are the foundry departments, grey iron, malleable, and ductile. Of the 586 hourly employees in these departments as of September 1973, 551 or 94 percent were black.21
As we mentioned, the incentive pay system at Stockham applies to highly repetitive jobs.22 Stockham’s production manager, Jack Marsh, in describing the work of an employee on the incentive system stated: “[H]e does the same thing over and over.” An employee receives incentive pay only if his production output exceeds the norm established by the company for the particular job. For this reason the pressures associated with incentive work at Stockham have led employees to call it “the racetrack”23 *328As of September 2, 1973, 70 percent of all black workers were assigned to incentive jobs as compared with only 31.7 percent of white employees. This concentration of blacks in incentive jobs and in the foundry departments supports the inference of discriminatory job assignments.
The pattern of job assignments at Stock-ham appears to result from a process that is largely subjective. First, employees are selected for individual seniority department jobs by departmental supervisors. The selection decision is totally discretionary and is not guided by written instructions.24 Second, apart from testing and seniority requirements, the decision to promote or transfer an employee into a new department is discretionary with the appropriate departmental supervisor and there are no written standards for' the selection process.25 The supervisory staff at Stock-ham is composed overwhelmingly of whites,26 and there are no safeguards against racial bias in the selection process.
In sum, even on the eve of trial the defendant Stockham discriminated in allocating jobs on the basis of race. A definite pattern of intentional racial staffing is revealed by statistical evidence on the disparities in black and white representation in seniority departments, on the relatively few integrated jobs in 1973, on the concentration of blacks in the lower job classes of both incentive and non-incentive jobs, and on the wage disparities between blacks and whites. In addition, it seems clear the undesirable working conditions associated with the jobs to which a vast majority of blacks are assigned verify the contention that jobs are allocated on the basis of race. The statistics must be evaluated in light of the admitted total segregation of jobs at Stockham until 1965; the persistent segregation of facilities and programs at least until 1974; and the roles played by white supervisors in discretionary and subjective assignment, transfer, and promotion decisions. This evidence taken together establishes a prima facie case of intentional discrimination according to the amicus EEOC.
The plaintiff in an action under Title VII has the burden of establishing a prima facie case of discrimination in employment practices. That burden may be met with statistical proof when it reaches *329proportions comparable to those in this case. Wade v. Mississippi Cooperative Extension Service, 5 Cir. 1976, 528 F.2d 508, 516-17; United States v. Hayes International Corp. (“Hayes II”), 5 Cir. 1972, 456 F.2d 112, 120. See also Pettway v. American Cast Iron Pipe Co., 5 Cir. 1974, 494 F.2d 211, 225 n.34; United States v. Jacksonville Terminal Co., 5 Cir. 1971, 451 F.2d 418, 442, 446, cert. denied, 1972, 406 U.S. 906, 92 S.Ct. 1607, 31 L.Ed.2d 815. Indeed the Supreme Court has recently approved the use of statistical proof in establishing a prima facie case of racial discrimination in the trucking industry. “Statistics are equally competent in proving employment discrimination. . Statistics showing racial or ethnic imbalance are probative in a case such as this one only because such imbalance is often a telltale sign of purposeful discrimination; absent explanation, it is ordinarily to be expected that nondiscriminatory hiring practices will in time result in a work force more or less representative of the racial and ethnic composition of the population in the community from which employees are hired.” International Brotherhood of Teamsters v. United States, 431 U.S. 324, 339 and n.20, 97 S.Ct. 1843, 1856, 52 L.Ed.2d 396 (1977) (Teamsters).
Here the plaintiffs have produced evidence of gross disparities in job allocations at Stockham on the basis of race. All but two of the seniority departments were either predominantly white or predominantly black at Stockham at the time of trial in 1973. Only sixteen percent of the hourly jobs were integrated by that time.27 In 1973 the overwhelming majority of both incentive and non-incentive white workers were employed in jobs with the highest job classifications.
Blacks earn, on the average, $0.37 less per hour than whites, including overtime and incentive pay.28 Seventy percent of all black employees work in the monotonous, pressurized conditions of the incentive system, and 94 percent of all workers subject to the hot, dusty, dirty conditions of the foundry departments are black. The disparities revealed by the statistics on job allocations at Stockham are gross and the statistical evidence compelling; they establish a clear prima facie ease of purposeful discrimination.
The statistical patterns do not complete the plaintiffs’ case. In addition, they offer persuasive evidence of total job segregation prior to 1965 and the intransigent retention of segregated facilities and programs at Stockham until at least 1974. As this Court recently observed through Judge Clark in Swint v. Pullman-Standard, 5 Cir. 1976, 539 F.2d 77, 97:
[T]he prior history of discriminatory job class assignments is clearly relevant to the issue of whether the present discrepancies in departmental assignments were part and parcel of a broad scheme to treat black and white workers differently. Historical policies of racial discrimination have often been used by other courts as indicia of plant-wide discriminatory conduct. (Footnote omitted.)
In erroneously concluding that Stockham has never discriminated in job assignments the district court did not have a finding of prior discriminatory job class assignments at Stockham on which to rely in evaluating the post-1965 statistical evidence. Nevertheless, the court had before it unrebutted evidence of post-Act policies of segregation in facilities and programs as indicia of discriminatory conduct on the part of Stock-*330ham. The district court erred in not relying on that evidence in evaluating the plaintiffs’ case.
In Bolton v. Murray Envelope Corp., 5 Cir. 1974, 493 F.2d 191, 195, this Court observed that the significance of statistical disparities between the races revealed in evidence of job assignment and employee discharges
is magnified when appraised in light of the fact that [the defendant] has little, if any, initial job qualifications requirements. In nearly every situation, the hiring, initial job assignment, and promotion is almost exclusively a subjective determination made by white supervisors.
Here initial job assignments were made on the basis of decisions by white departmental supervisors29 without any formal selection standards or written guidelines. Similarly, transfer decisions were made without objective standards by the largely white supervisory staff. For promotions, seniority controlled only when the supervisor decided that applicants were approximately equal in qualifications, a largely subjective decision made by predominantly white supervisors. In Rowe v. General Motors Corp., 5 Cir. 1972, 457 F.2d 348, 359, this Court examined procedures involving subjective evaluations of employees:
All we do today is recognize that promotion/transfer procedures which depend almost entirely upon the subjective evaluation and favorable recommendation of the immediate foreman are a ready mechanism for discrimination . . . . We and others have expressed a skepticism that black persons dependent directly on decisive recommendations from Whites can expect non-discriminatory action.
See also Pettway v. American Cast Iron Pipe Co., 494 F.2d at 231-32; United States v. Jacksonville Terminal Co., 451 F.2d at 442. In addition, in Pettway this Court emphasized that “[c]ourts have condemned procedures for promotion and job assignment which are not objective and uniform.” Pettway v. American Cast Iron Pipe Co., 494 F.2d at 232 n.47.
We conclude that the plaintiffs’ evidence of racial disparities in job allocations after 1965, of total job segregation prior to 1965, of the continued segregation of facilities and programs until long after the effective date of Title VII, and of subjective assignment, transfer, and promotion decisions by white supervisors at Stockham make a prima facie case of discriminatory employment practices in job allocations at Stockham. Once the plaintiff in a Title VII case has presented a prima facie case of discrimination, the “onus of going forward with the evidence and the burden of persuasion” is on the defendant. United States v. Hayes International Corp., 456 F.2d at 120. The strength of the evidence presented in this case imposes on the defendant a heavy burden in attempting to counter the inference of systematic and purposeful discrimination.
The defendant Stockham seeks to rebut any “colorable” inference of discrimination primarily by means of two arguments, both of which focus on the disparities in job classes of positions in which blacks and whites are employed. First, Stockham contends that an employee’s job class has only a negligible impact upon employee earnings. In support of this argument the defendant points to four facts: (a) that incentive workers earn approximately 25 percent more than their base pay rate, (b) that of 75 employees in the production and maintenance unit making more than $9,000 a year in 1972, 40 of these persons were in job classes 9 and below, (c) that no JC 11 or 12 workers earned more than $9,000 in 1972, and (d) that some black employees earned more than white employees. The district court relied on these same facts in finding that Stockham has not engaged in discriminatory job assignments. These four facts are unpersuasive of the “negligible” impact of job class on employee earnings and large*331ly irrelevant to the issue of present discrimination in job allocations. First, the evidence is uncontradicted that the job class in which a job is located determines the base pay rate for the job and that the lower the job class, the lower the pay rate. That incentive workers earn more than the base rates for their jobs does not refute the direct relationship between pay rate and job class. Further, that some incentive workers are industrious enough to earn more than $9,000 also does not refute the relationship. Moreover, that workers in JC 11 and 12 do not earn more than $9,000 suggests only that incentive pay augments base pay. More importantly, it suggests that incentive jobs may not be as desirable as the defendant would have us believe.30 Workers obviously seek the skilled craft jobs in job classes 11 and 12 even though they are not on the incentive system and thus have built-in limitations in pay. In its argument the defendant overlooks the fact that many jobs in JC 11 and 12 serve as training grounds for JC 13 positions and thus'have greater earnings potential over the long term than do incentive jobs with lower job classes.
Finally, Stockham attempts to show that whites in one of the higher-classified jobs, crane operator in the iron grey foundry, did not earn more than some of the blacks in that department working in lower job classifications. For example, as the plaintiffs point out, one white crane operator in 1972 earned less than 81 of the 292 blacks in the department. When the data are controlled for seniority, however, and the white crane operator’s salary is compared with that of blacks hired the same year as he was, 1968, the evidence is more revealing. The white employee had a higher earning rate per hour than all but one black.
The defendant seeks to refute the plaintiffs’ evidence that blacks earn an average of $0.37 less per hour than whites, including incentive and overtime pay, by means of the testimony of an expert witness. Dr. James Gwartney, an economist who testified for the defendant, conducted a study of the earnings of production and maintenance employees at Stockham in an attempt to determine the factors that explain earnings disparities between employees. He concluded that such productivity factors as education, skill, building experience, craft skill level, and absenteeism — not discrimination — explain the earnings differences between blacks and whites at Stockham. In studying the earnings opportunities at Stockham Dr. Gwartney considered four factors: (1) the earnings of employees at Stockham compared with the earnings of those in local, regional, and national labor markets and with earnings in other companies; (2) relative changes in the earnings of company employees over a long period of time; (3) relative changes in the earnings of company employees recently hired; and (4) application of the residual approach of scientifically adjusting earnings for productivity factors.
Dr. Gwartney’s analysis does not meet the point that wage differences between blacks and whites at Stockham are explained by racially discriminatory job allocations. The first three factors are irrelevant to the question of discrimination at Stockham. The critical question is whether blacks at Stockham earn less than whites at Stockham, not whether blacks at Stockham earn more or less than blacks in various other geographic areas or in other companies. Those statistics will suggest only whether there is more or less discrimination in earnings opportunities for blacks in other settings as compared with Stockham. In addition, such statistics are totally irrelevant to the issue whether blacks are segregated by jobs and departments at Stockham and to the issue whether blacks must earn their wages under conditions less desirable than those of whites.
*332On its face, Dr. Gwartney’s fourth factor deals with relevant and persuasive statistics on earnings disparities between blacks and whites. His regression analysis of productivity factors will not stand scrutiny.
Regression analysis is a statistical method that permits analysis of a group of variables simultaneously as part of an attempt to explain a particular phenomenon, such as earnings disparities between blacks and whites. The method attempts to isolate the effects of various factors on the phenomenon. Dr. Gwartney’s analysis is based on the assumption that productivity factors, not discrimination, may explain the wage differences between Stockham’s black and white employees. The productivity factors Dr. Gwartney employed were years of schooling, achievement, seniority, skill level, outside craft experience, outside operative experience, absenteeism, and merit ratings.
The rub comes with how these factors were defined in Dr. Gwartney’s study. As the plaintiffs point out, the critical factors of “skill level” and “merit rating” were defined in such a way as to incorporate discrimination. “Skill level” was derived from an employee’s job class; he had “skill” only if he worked in a job with a rating between JC 10 and 13. The systematic exclusion of blacks from promotion and training opportunities for such jobs, as is alleged here, will automatically produce no black employees with “skill level”. A regression analysis defining “skill level” in that way thus may confirm the existence of employment discrimination practices that result in higher earnings for whites.
Dr. Gwartney used the merit ratings of Stockham supervisors, who are overwhelmingly white, for his “merit rating” factor; blacks average 71.3 in these ratings while whites average 79.3. If there is racial bias in the subjective evaluations of white supervisors, then that bias will be injected into Dr. Gwartney’s earnings analysis.31
Further, Dr. Gwartney included education as one of his productivity factors, even though education is not a job requirement at Stockham, because, according to the defendant, “an individual’s educational level, regardless of race, impacts earnings”. The fallacy in this conclusion stems from two facts: (1) as the defendant concedes, education is not a job requirement at Stockham, and (2) white employees at Stockham have more education than blacks.32 Thus, adjusting for education in a regression analysis of earnings where education is not related to job performance and where one race is more educationally disadvantaged than another, masks racial differences in earnings that may be explainable on the basis of discrimination. Certainly such differences cannot fairly be explained on the basis, of a factor, such as education, concededly irrelevant to adequate job performance.33
Significantly, although Dr. Gwartney asserted that his study proves that productivity factors and not discrimination explain the wage differences between black and white employees at Stockham, he concedes that he made no attempt to control or check for racial bias in his analysis. Our examination of his analytical approach compels us to conclude that the results of Dr. Gwartney’s study in no way refute the plaintiffs’ prima facie case of racial discrimination in job allocations at Stockham.
Stockham’s attempt to refute the plaintiffs’ evidence of racial job allocations by focusing on earnings differences misses the *333point. First, such an emphasis ignores the lopsided statistics on the number of all-black and all-white jobs at Stockham. Second, the defendant’s focus on earnings avoids consideration of whether job segregation by itself, apart from any issue of economic harm, violates Title VII. This Court recently ruled on this issue in Swint v. Pullman-Standard, 539 F.2d at 89-90, in an opinion by Judge Clark:
[A] Title VII plaintiff does not have to show economic loss to prove discrimination.
The key for this case is whether there was past discrimination . . Going further and requiring plaintiffs to prove that past assignment practices produced lower pay checks is contrary to law and precedent. .
Title VII contains neither requirement nor implication that economic harm must be shown before a class can be found to have made out a prima facie case of racially discriminatory job assignment. Indeed, the statutory prohibitions of the enactment are explicitly broader than economic harm.
Thus, not only is the defendant’s attempt to rebut the inferences of discrimination presented by the plaintiffs’ evidence factually inadequate, it is also legally insufficient.
The defendant seeks to augment its rebuttal of the plaintiffs’ prima facie case with a second argument on the issue of the working conditions associated with departments and jobs in which blacks work. The defendant relies in part on the finding of the district court that the plaintiffs’ allegation that blacks work in the hottest, dirtiest, and dustiest jobs at Stockham is unsupported. We have previously discussed the evidence offered by the plaintiffs to substantiate this assertion.34 We reject the conclusion that the allegation is unsupported.
In addition, the defendant contends that the plaintiffs failed to establish any correspondence between job classifications and work conditions. Stockham argues that many of the job class 10 through 13 positions held at Stockham by whites involve working conditions similar to the ones described by the plaintiffs. This argument does not reach the evidence that an overwhelming majority of blacks work in the tedious and pressure-filled atmosphere of incentive jobs while a substantial majority of whites do not. In addition, this argument in effect confirms another of the plaintiffs’ contentions, that even in the largely all-black foundries the few jobs held by whites are the ones in the highest job classes. Even more significantly, this defense, focusing on the relative desirability of jobs and departments, like the defendant’s emphasis on earnings data, is legally irrelevant. In Swint we made clear that departmental desirability is not an essential part of a plaintiff’s prima facie proof. Swint v. Pullman-Standard, 539 F.2d at 91. In Reed v. Arlington Hotel Co., Inc., 8 Cir. 1973, 476 F.2d 721, 723, discussed in Swint, the Eighth Circuit concluded: “[Statistics which show segregated departments and job classifications establish a violation of Title VII.” Along with that court, our concern is that black employees have suffered “the indignities of segregation”. Id. at 726. Here the plaintiff presents undeniable evidence of segregated jobs, the concentration of blacks in certain departments, the lengthy unlawful segregation of facilities and programs, the admitted total allocation of jobs on the basis of race before 1965, and the subjective selection of employees for assignment, transfer, and promotion by an overwhelmingly white supervisory staff. Evidence of disparities in the earnings and working conditions of blacks and whites are persuasive, as in this case, but unnecessary to a determination that Title VII has been violated in the allocation of jobs on the basis of race.
*334The district court bases its holding that Stockham did not discriminate in job assignments after 1965 essentially on two conclusions: (1) that blacks were not qualified for more skilled positions and (2) that blacks voluntarily chose the jobs to which they were assigned. The EEOC points out the dearth of factual support for the court’s reasons. First, the court relies on the testimony of Flount R. Hammock, manager of the Alabama State Employment Service in Birmingham, in support of its first conclusion. But Hammock stated: “[W]hen we’re talking about skilled trade, there are very few white or black available.” Thus the court’s first factual conclusion is clearly erroneous. In addition, Stockham trains a substantial number of all craftsmen it employs at Stockham.
In support of its conclusion that blacks were assigned to positions they requested or preferred, the district judge stated:
Of the 626 employees (251 white and 375 black) employed as of January 1, 1974, who had sought a specific job when applying to Stockham for employment, 61% of the white employees and 53% of the black employees were placed in the job of their own selection.
394 F.Supp. at 455. As the EEOC correctly observes, these statistics are factually insufficient and legally irrelevant. First, over two-thirds of Stockham’s work force has been assigned to jobs without assignment requests. In addition, these statistics do not reflect the number of blacks not hired at Stockham who sought traditionally white jobs and were rejected. Further, the statistics do not include information on the number of blacks who sought to transfer to jobs in traditionally white departments or positions after having worked for a time at Stockham. Finally, this Court has frequently recognized the “meaningless request” phenomenon. Pettway v. American Cast Iron Pipe Co., 494 F.2d at 232; Bing v. Roadway Express, Inc., 5 Cir. 1973, 485 F.2d 441, 451. Recently, in Teamsters, 431 U.S. at 367, 97 S.Ct. at 1870, the Supreme Court observed:
[A] nonapplicant can be a victim of unlawful discrimination . . . when an application would have been a useless act serving only to confirm a discriminatee’s knowledge that the job he wanted was unavailable to him.
The district court’s reliance on the percentage of requests for specific jobs granted new black employees is legally misplaced in cases such as the one before us.
We hold that the district court erred in concluding that jobs have never been allocated on the basis of race at Stockham. There is overwhelming evidence in the record of past and present discrimination in the allocation of jobs in violation of Title VII.
C. Testing Program
1. Wonderlic Personnel Test
Although only the Bennett Mechanical Comprehension Test for screening apprentice program candidates was in use before 1965, in August of that year, just after Title VII became effective, Stockham instituted an extensive testing program. The company adopted the Wonderlic Personnel Test for use in screening virtually all applicants for hourly jobs at Stockham, new hirees, employees seeking intra- and interdepartmental transfers and promotions, and individuals requesting selection for the apprentice program. The Wonderlic Test was employed at Stockham as a screening device until 1971.
The plaintiffs say that the district court erred in failing to find that Stockham’s Wonderlic testing program was unlawful. The district court found:
Plaintiffs have not offered any competent evidence, direct or indirect that the Wonderlic test, as administered at Stock-ham in the period from 1965 through 1971, disqualified a disproportionate num*335ber of black applicants or employees. In the absence of evidence of a racially disproportionate impact, the defendant-employer was under no obligation to go forward and prove that the Wonderlic test, as administered during that period, was job-related.
394 F.Supp. at 498. The defendant concedes that the Wonderlic Test was never validated. Stockham did not offer evidence of the job-relatedness of the test, but as the Supreme Court observed in Albemarle Paper Co. v. Moody, 1975, 422 U.S. 405,425, 95 S.Ct. 2362, 2375, 45 L.Ed.2d 280:
This burden arises, of course, only after the complaining party or class has made out a prima facie case of discrimination, i. e., has shown that the tests in question select applicants for hire or promotion in a racial pattern significantly different from that of the pool of applicants.
The primary issue here in relation to the Wonderlic Test is whether the plaintiffs produced sufficient evidence of the adverse impact of the test on black employees at Stockham.
The plaintiffs’ evidence on the racial impact of the Wonderlic Test consists of five elements. First, on cross-examination Dr. Joan Haworth, the defendant’s statistician, stated that in a computer study she conducted at Stockham’s request she found that black employees at Stockham scored lower than white employees.35 Second, E. F. Wonderlic & Associates, the developers and publishers of the Wonderlic Test, undertook a nationwide study on “Negro norms” and concluded that over a wide range of educational levels, jobs, and geographic areas there is a “stable differential” in black and white scores on the test with blacks scoring approximately eight points lower than whites.36 See 394 F.Supp. at 482. Third, the plaintiffs produced expert testimony that blacks tend to perform less well on general intelligence tests, such as the Wonderlic, than whites do. Fourth, the Wonderlic Test has been condemned in a number of cases for having an adverse impact on blacks.37 Fifth, the plaintiffs’ evidence showed that disproportionately few blacks were selected for jobs or training programs for which high Wonderlic scores were required.38
*336The district court rejected the plaintiffs’ evidence on the adverse impact of the Wonderlic Test for a variety of reasons. See 394 F.Supp. at 482-83. The court’s principal reason for this finding was that the plaintiffs failed to present any evidence “to the effect that the average score of blacks at Stockham on the Wonderlic Test was any lower from the average score of whites” except for the testimony of Dr. Haworth, which the court concluded was incomplete and therefore insufficient. In addition, the court found:
Plaintiffs failed to relate any supposed difference in test performance by blacks and whites to the Wonderlic cutoff scores actually utilized by Stockham from 1965 to 1971.
394 F.Supp. at 483.
The court gave too much weight to the plaintiffs’ failure to offer evidence on the actual scores black and white employees at Stockham achieved on the Wonderlic Test.39 In Pettway v. American Cast Iron Pipe Co., 494 F.2d at 221, we concluded that the plaintiffs had shown that testing and educational requirements had an adverse impact on black employees seeking promotions with evidence that fewer black employees were promoted between 1965 and 1971. Here the plaintiffs presented substantial statistical data on the gross disparities between the number of black and white workers in jobs requiring higher Wonderlic scores.40 This Court recognized in Watkins v. Scott Paper Co., 530 F.2d at 1185:
[A] statistical showing of black exclusion from a particular kind of job establishes a prima facie ease of discrimination.
See also Vulcan Society of the New York City Fire Department, Inc. v. Civil Service Commission, 2 Cir. 1973, 490 F.2d 387, 393, *337in which Judge Friendly rejected any requirement that the racially disproportionate impact of employment tests be proved with “complete mathematical certainty”.
The plaintiffs’ statistical evidence shows that in three kinds of positions for which high Wonderlic scores were required black employees were greatly underrepresented. Although 66 percent of all production and maintenance workers were black, in November 1970, only 4 percent of employees working in jobs with a JC 9 through 13 rating were black.41 In addition, as we have already observed,42 the statistics show that 96 percent of all white employees at Stockham were in jobs requiring a score of 15 or higher while only 13 percent of the black workers were in such jobs. Even more strikingly, there were no black apprentices in March 1971 when use of the Wonderlic Test was discontinued. Finally, 56 percent of the entire work force at Stockham is black although only 7 percent of the clerical workers were black in 1971. These statistics show that during the period the Wonderlic Test was administered blacks were totally excluded from the apprenticeship program, were substantially excluded from clerical jobs, and were disproportionately excluded from jobs classified in JC 9 through 13.
This evidence, considered with Dr. Ha-worth’s general finding that blacks scored lower on the Wonderlic Test at Stockham, the results of the “Negro norms” study by Wonderlic & Associates, and the expert testimony that general intelligence tests such as the Wonderlic generally have disparate effect on blacks establishes that the plaintiffs met their initial burden of showing that the Wonderlic Personnel Test had ah adverse impact on black employees at Stockham. The district court erred in reaching the contrary conclusion. Given the adverse effect, the burden of persuasion shifted to the company to show the job relatedness of the test. Pettway v. American Cast Iron Pipe Co., 494 F.2d at 222. Stockham offered no evidence to satisfy that burden. As a result, the plaintiffs are entitled to equitable relief.43
*338Alternatively, apart from the issue whether the test itself had a discriminatory effect on blacks, the plaintiffs and the amicus EEOC argue that the way in which the Wonderlic test was administered with a dual scoring system for inter- and intradepartmental transfers, constituted an “artificial, arbitrary, and unnecessary” barrier to employment opportunities condemned by the Supreme Court in Griggs v. Duke Power Co., 401 U.S. at 431, 91 S.Ct. 849. As we have already discussed,44 Stockham required employees transferring to jobs within their own departments to achieve the “minimum” score designated for the job; whereas, employees transferring interdepartmentally were required to achieve the higher “norm” score for the position. As the EEOC points out, the effect was that black employees who had been excluded from certain departments on account of race were required to score higher on the screening test than whites who never suffered such discrimination. Thus, this requirement tended to restrict blacks to departments to which they had been discriminatorily assigned. The requirement was arbitrary, in that an employee who achieved the “minimum” score was deemed qualified to perform the job. Further, if the purpose of the requirement was to adjust for the greater intradepartmental experience of employees already working in the department, that rationale is questionable. Stock-ham has no lines of progression among jobs; hence, experience in one job is not seen as necessary for satisfactory performance in another. Intradepartmental experience could have been considered as a separate criterion with more accurate results than with an adjustment to scoring on a general intelligence test.
Finally, the company apparently recognized that there was only a minimal relationship between satisfactory job performance and high scoring on the Wonderlic Personnel Test; an employee performing well within a department was not required to achieve a high score to be selected for another job in that department. This fact may explain why Stockham did not attempt to show the job-relatedness of the Wonderlic test. This Court’s conclusion in Pettway on a preferential bidding system applies equally well here:
The requirement that bids from within the craft departments be given initial, primary consideration must fail in light of the proof demonstrating that a large majority of black employees have been excluded from these departments.
Pettway v. American Cast Iron Pipe Co., 494 F.2d at 240. Therefore, the dual scoring system for the Wonderlic test, giving preference to employees transferring within departments, was unlawful even apart from the issue of the unlawfulness of the test itself.
2. Tabaka Testing Program
The district court found that the Tabaka tests had been in use at Stockham since July 17,1973. The court also concluded:
The Tabaka tests are actively considered in selection decisions; however, no employee has been disqualified on the basis of the Tabaka tests, even though some have failed to attain the failure probability score.
394 F.Supp. at 486. The plaintiffs, however, say that they were stymied in their efforts to collect information on whether the Tabaka tests adversely affect blacks because of the emphatic assertion of Stock-*339ham’s employee testing manager that the Tabaka tests have not been used in any employment decisions.
The district court’s finding that the Tabaka tests have been in use since July 17, 1973, is apparently based on the testimony of E. Reeves Sims, the company employee relations manager. We have concluded that the factual finding is clearly erroneous. Jack H. Adamson, whose direct responsibilities include Stockham’s testing program, testified that the tests were administered only for the purpose of collecting data on their effect and not as part of employee selection decisions.45 Therefore, we do not reach the question whether Stockham has succeeded in proving the job relatedness of the Tabaka tests. Instead, we remand that issue to the district court so that additional evidence can be taken. In more than three years since the trial, the company must have accumulated a substantial quantity of data. Such statistical information should be sufficient both for a determination of whether the Tabaka tests have an adverse impact on blacks and for augmentation of the validation study.
If the district court finds on remand that the Tabaka test has an adverse impact on blacks, then it should evaluate the defendant’s proof of job relatedness in light of the Supreme Court’s opinion in Albemarle Paper Co. v. Moody, 422 U.S. 405, 95 S.Ct. 2362, 45 L.Ed.2d 280. In addition, if the defendant continues to rely on the Tabaka concurrent criterion study as the basis for its assertion that the tests are job-related, the district court should consider whether the job analysis conducted by Victor Tabaka meets EEOC requirements specified in 29 C.F.R. § 1607.5(b)(3) and (4) and conforms with the professional standards of the American Psychological Association (“APA”). In United States v. City of Chicago, 7 Cir. 1977, 549 F.2d 415, 431, the Seventh Circuit applied the EEOC Guidelines and quoted the APA standards in emphasizing the importance of the quality of criteria chosen for criterion-related validity studies:
[T]he logic of criterion-related validity assumes that the criterion possess validity. All too often, tests are validated against any available criterion with no *340corresponding investigation of the criterion itself. The merit of a criterion-related validity study depends on the appropriateness and the quality of the criterion chosen . . . . Criterion-related validity studies based on the ‘criterion at hand’ chosen more for availability than for a place in a carefully reasoned hypothesis, are to be deplored.
As that court concluded, “The entire rationale of a criterion-related study requires that the criterion with which the test results are compared be a good measure of job performance.” Id. at 433.
Finally, in the event the district court should conclude that the Tabaka tests are job-related, the court should permit the plaintiffs to present evidence that alternative selection devices, without similarly undesirable racial effect, are available to the company. Albemarle Paper Co. v. Moody, 422 U.S. at 436, 95 S.Ct. 2362. We remand the issue of the lawfulness of the Tabaka tests to the district court.
D. Selection and Training of Craftsmen
The plaintiffs and EEOC insist that the district court erred in holding that Stockham has never discriminated against blacks in the selection and training of craftsmen. In essence they say that the company’s practices with regard to craft positions contain elements of present discrimination and “neutral” policies which perpetuate past bias in hiring.
First, the plaintiffs contend that the same discriminatory practices that denied blacks particular jobs and positions in certain departments have denied blacks training and advancement to craft positions. They point to evidence of the gross statistical disparities in the number of black and white craftsmen and the total exclusion of blacks from craft training programs until April, 1971. Second, the plaintiffs and the amicus point to the screening requirements used by Stockham in selecting employees for the apprenticeship training program. Since 1953 a passing score on the Bennett Mechanical Test has been required. Between 1965 and 1971 a score of 18 on the Wonderlic Personnel Test was required. Since 1970 an applicant for apprentice training has been required to have a high school diploma, or its equivalent, and to be thirty years old or younger, excluding time in military service. Finally, the plaintiffs stress the total discretion of an overwhelmingly white supervisory staff in selecting employees for the apprenticeship program and for on-the-job training.
In response to these contentions the defendants emphasize the trial court’s finding that the plaintiffs failed to show a single instance involving the company’s failure to select a qualified black craftsman in favor of an equally or less qualified white. In addition, the defendant stresses the differences in skill level and productivity between black and white employees allegedly revealed by the regression analysis of earnings differences in the study conducted by Dr. Gwartney. We have already discussed the Gwartney study and rejected it as a basis for rebutting the prima facie case of discrimination established by the plaintiffs’ evidence.46 That analysis applies equally well here. The defendant’s first argument, that the plaintiffs have failed to show that a qualified black has been rejected for craft training, is legally irrelevant. In Rowe v. General Motors, supra, 457 F.2d at 355 n.14, we explained how the burden of proof is allocated in Title VII cases:
The Trial Court took the simple position that the employees failed to carry the burden of establishing that GMAD maintained a policy of discrimination in employment practices . . . . [With Griggs] the Supreme Court has made clear that the employer has the burden of showing that any given requirement which has a tendency to reduce job opportunity because of race has a demonstrable *341relationship to the employment in question. Griggs, supra, 401 U.S. at 431, 91 S.Ct. [849] at 853, 28 L.Ed.2d at 164.
Thus, the trial court here concluded that the plaintiffs failed to meet their burden of proof under an erroneous view of controlling legal principles.47
We must examine the requirements and procedures for selecting employees into craft training programs to determine whether they operate to discriminate against blacks.
1. Representation of Blacks in Craft Jobs and Training Programs.
The company defines craft jobs as those in job classes 9 through 13. As to the representation of blacks in craft positions at the time of trial the district court found:
Black employees fill 10 (5%) of approximately 200 craft jobs at Stockham. Five percent is not an underrepresentation of blacks in craft jobs compared to the local and national labor markets.
394 F.Supp. at 455. The court’s conclusion that a five percent representation of blacks in craft jobs is not disproportionate because it compares favorably with local and national labor markets is clearly erroneous. The relevant work-force for comparison purposes is Stockham where 66 percent of all maintenance and production workers are black.48 When compared with that figure, 5 percent looks paltry indeed. In fact, statistical evidence shows that in June 1973, while only 6 blacks were working in craft jobs at Stockham, 227 whites were employed as craftsmen. Further, in June 1965 there were 70 white craftsmen and no blacks; similarly in June 1968 there were no black craftsmen and 85 whites in craft jobs.
In addition, the court cited as evidence that Stockham does not discriminate against blacks with regard to craft positions the fact that no black employee who has completed the apprentice program at Stock-ham or elsewhere is not employed in a craft position. See 394 F.Supp. at 455. This finding fails to prove a lack of discrimination because it ignores three facts: (a) the testimony of company officials that a substantial number of craftsmen at Stockham have been trained in the company’s apprenticeship program;49 (b) that no black was admitted to the apprenticeship training program until April 1971; and (c) that only 6 of 101 employees selected by the company for apprenticeship training since July 2, 1965, have been black.
The district court however found that the apprentice program has never been restricted to whites. 394 F.Supp. at 475. As evidence for this conclusion the court cited the following statistics:
From 1965 through 1973, a total of 65 timely applications for apprentice positions were filed, 14 by black and 51 by white employees. Of that total 38 were granted, 6 for black and 32 for white employees.
394 F.Supp. at 477. These statistics do not serve factually or legally to counter the inference of discrimination raised by the plaintiffs’ statistics on black participation in the apprentice program. The filing of a timely application has never been a requirement for admission to the apprentice program. Of 101 employees admitted to the program since 1965, 63 did not file timely applications. More significantly, every employee who entered the program without filing a timely application was white. Thus, the court’s citation of statistics to show that a comparable number of timely *342applications to the apprenticeship program were granted for black and white employees is largely irrelevant. Further, no black filed an application for the program until 1971. Such applications were apparently seen by black employees as “useless acts”, Teamsters, 431 U.S. at 368, 97 S.Ct. 1843, because of the total exclusion of blacks from the program until 1971. Thus, the court’s reliance on the number of timely applications filed and granted by race to prove a lack of discrimination in apprentice training was erroneous. As we asked in another case, “[i]f an employee realizes full well that blacks simply are not hired [for certain positions], why should he bother to apply?” Bing v. Roadway Express, Inc., 485 F.2d at 451. Just recently the Supreme Court recognized, as no new principle, that unlawful employment practices may be so successful as to totally deter victims of gross and pervasive discrimination from applying for jobs. Teamsters, 431 U.S. at 368, 97 S.Ct. 1843.
We conclude that the district court was clearly erroneous in finding that Stockham
has never had a policy of excluding blacks from craft jobs and the apprenticeship program.50 The next question is whether the procedures established by Stockham for screening and selecting apprentices served to accentuate and to perpetuate the effects of such discrimination.
2. Testing Requirements
Since 1953 applicants for apprentice training have been required to achieve a passing score on the Bennett Mechanical Test. The district court found that the Bennett test did not have an adverse impact on blacks because beginning in 1969 or 1970 the company did not require black employees to attain the same scores on the test as whites. See 394 F.Supp. at 483. That finding is clearly erroneous. Edward Glenn, the Stockham manager in charge of administering and grading that test in 1969 and 1970, testified without qualification that the examinations completed by black and white employees were graded the same way and the results were treated equally. He denied that a dual scoring system had *343been employed. That no blacks were admitted to the apprenticeship program before 1971 raises a prima facie case of discrimination. Watkins v. Scott Paper Co., 530 F.2d at 1185-86. The defendant Stock-ham failed to present evidence of the job-relatedness of the test. Therefore, the use of the Bennett test for selection of apprentices was unlawful.
The Wonderlic Personnel Test was used for screening employees seeking admission to the apprenticeship program from August 1965 until March 1971. Stockham required a score of 18 on that test. In subsection “III.B.” of this opinion we discussed the lawfulness of the Wonderlic testing program and concluded that the plaintiffs made a prima facie case of the adverse impact of the test on blacks seeking jobs, including apprentice training, for which higher scores were required. As we stated, because the defendant offered no evidence of job-relatedness, we have concluded that the plaintiffs are entitled to equitable relief from the effects of the Wonderlic testing program.
3. Educational Requirements
In 1970 Stockham instituted a requirement that all apprentices have a high school diploma or its equivalent. Before 1970 only a grammar school education was required. Under Title VII, practices and procedures “cannot be maintained if they operate to ‘freeze’ the status quo of prior discriminatory employment practices”. Griggs v. Duke Power Co., 401 U.S. at 430, 91 S.Ct. at 853. “If an employment practice which operates to exclude Negroes cannot be shown to be related to job performance, the practice is prohibited.” Id. at 431, 91 S.Ct. at 853. The plaintiffs contend that the high school education requirement is a “neutral” employment practice that operates to perpetuate the effects of past discrimination because it has an adverse impact on black employees and has not been shown to be job-related. The district court reached a contrary conclusion, finding:
There has been no showing on the part of plaintiffs that the education and age guidelines produce a disproportionate impact on black applicants, ... as there is no significant difference between the percentage of black and white employees at Stockham possessing high school educations. Until a showing of disproportionate impact is made, Stock-ham is not required to demonstrate that the education and age requirements are job related. (Footnotes omitted.)
394 F.Supp. at 497. The evidence shows that 61.5 percent of the white hourly workers have a high school education; whereas, only 50.1 percent of the black hourly workers meet that requirement. Because 66 percent of the production and maintenance workers are black, these percentages mean that 280 white employees as compared with 772 blacks are disqualified from the apprenticeship program by this requirement. Further, while 58 white workers have been selected for apprentice training since the requirement was imposed in 1970, only 6 black employees have been chosen. Apparently as a second reason for finding there was no adverse impact on blacks from the operation of the education requirement, the district court found that the requirement is not automatically applied, citing the following statistics:
Since 1965 the high school education level requirement for the apprentice program has been waived on 4 occasions, 3 times for white employees and once for a black employee.
394 F.Supp. at 477. The statistics cited by the court fall of their own weight. While fewer blacks have a high school diploma than whites, the requirement has been waived for only four employees out of 64; three of these were white.
The evidence that fewer blacks than whites proportionately have attained a high school education and that only 9 percent of the employees chosen for the program during the operation of the educational requirement have been black when two-thirds *344of the work force is black compels us to conclude that the high school requirement is discriminatory in its effect. Therefore, Stockham had the burden of showing that the requirement has “a manifest relationship” to the apprenticeship program. Griggs v. Duke Power Co., supra, 401 U.S. at 432, 91 S.Ct. at 854. The defendant offered no evidence to satisfy this burden.
4. Age Requirement
In 1970, at the same time that Stockham imposed the high school education requirement for selection to the apprenticeship program, the company added a maximum age limit for apprenticeship training of thirty, excluding time in military service. The plaintiffs cite this as another requirement that serves to “freeze in” the effects of past discrimination. As we cited in the preceding section of this opinion, the district court found that the requirement had no disproportionate effect on black employees. 394 F.Supp. at 497. We do not agree.
Because of the company’s policy of excluding blacks from the apprenticeship program in the past, the age requirement operates to prevent all blacks who reached the age of thirty before 1971, the first year any blacks were admitted into the program, from having an opportunity for apprenticeship training. In comparison, no white has been excluded by the age requirement who was not subject to consideration for the program before he became thirty years of age.
The age requirement thus operates as a practice neutral on its face but perpetuating the effects of past discrimination. For such a practice to be lawful under Title VII, it must be related to job performance or the employer must show that the requirement is a business necessity. As we observed in Local 189, United Papermakers and Paperworkers v. United States, 5 Cir. 1969, 416 F.2d 980, 989:
When an employer or union has discriminated in the past and when its present policies renew or exaggerate discriminatory effects, those policies must yield, unless there is an overriding legitimate, nonracial business purpose.
The company contends that the age limit is necessary to protect its investment in the lengthy apprenticeship training program. This justification, without more, cannot meet the burden imposed by the doctrine. The business necessity of a practice is not shown merely with evidence that it serves “legitimate management functions”.
“Necessity connotes an irresistible demand. To be preserved, the seniority and transfer system must not only directly foster safety and efficiency of a plant, but also be essential to those goals. . If the legitimate ends of safety and efficiency can be served by a reasonably available alternative system with less discriminatory effects, then the present policies may not be continued.” United States v. Bethlehem Steel Corp., 2d Cir. 1971, 446 F.2d 652, 662 [1971].
United States v. Jacksonville Terminal Co., 451 F.2d at 451. The thirty-year maximum age limit has not been shown to be essential to the goals of safety and efficiency. Any lesser showing is insufficient. There is nothing particularly compelling about the company’s decision to protect its training investment when the employee is thirty. The unfairness becomes apparent when it is considered in view of Stockham’s imposition of the age requirement on apprenticeship applicants in 1970, before beginning to admit blacks to the program and when no whites had ever been excluded on the basis of age. See EEOC Guideline 29 C.F.R. § 1607.11 (1974) and footnote 43. The plaintiffs are entitled to equitable relief from the continued application of this requirement.51
*3455. Supervisory Discretion
The plaintiffs and the EEOC contend that the disproportionately small number of blacks selected for the apprenticeship program since 1971 is also caused by subjective, discretionary selection procedures. The evidence shows several critical facts. The foreman or superintendent of the department in which a craft vacancy occurs selects a candidate for the apprenticeship program; that recommendation is almost invariably followed by the apprenticeship committee, which makes the final selection. Supervisors in other departments may also recommend employees. There are no formal written guidelines for the selection decision. Only such general factors as “desire” and “aptitude” for the craft position are considered. Just five of 120 foremen and none of the 26 superintendents or six general foremen are black. Thus, the plaintiffs are complaining of a largely subjective selection process involving white supervisors.52 This procedure presents a “ready mechanism” for discrimination. Rowe v. General Motors Corp., 457 F.2d at 359. Such subjective procedures when combined with statistical evidence on the grossly disproportionate number of blacks selected for the apprenticeship program since 1971;53 the unvalidated testing, educational, and age requirements that adversely affect blacks and the continuation of segregated facilities and programs until 1974, make a conclusive showing of present discriminatory practices. In Pettway v. American Cast Iron Pipe Co., 494 F.2d at 239, this Court reached a similar determination:
The historical formal exclusion and the statistical and testimonial evidence demonstrating disproportionate exclusion of blacks by the testing and educational requirements, when combined with the continuing use of the high school education or its equivalent standard and the present age requirement and lengthy apprenticeship term, constitutes not merely a prima facie case, but conclusive proof of present effect from past discrimination. (Footnotes omitted.)
See also United States v. Jacksonville Terminal Co., 451 F.2d at 442. We think that conclusion is equally applicable here.
E. Selection of Supervisors
The plaintiffs and the EEOC contend that the district court’s conclusion that the company did not discriminate in the selection of supervisors is clearly erroneous. In support of this contention they point to (1) the statistical evidence of a gross disparity between the number of blacks working in supervisory positions at Stockham and the number of black production and maintenance employees, and (2) the subjective selection procedures.
*346The evidence shows that there were no black supervisors at Stockham until May 1971. In 1973 of 120 foremen only 5 were black. There were no blacks of 26 superintendents and 6 general foremen. A majority of supervisors are selected from hourly employees of whom two-thirds are black. Some are chosen directly from the hourly work force and others are selected from the company’s personnel development program (“PDP”), which trains employees for supervisory positions.
According to Jack Marsh, the production manager at Stockham, the principal criterion for selecting an hourly employee as a supervisor is the “best man” for the job. He stated that such other qualifications as desire to be a foreman, work record, knowledge of the job, training, physical fitness, and common sense are also considered.
There are no formal, written guidelines for the selection process. The superintendent of a department having a vacancy in the supervisory staff selects candidates for the job and consults with the production manager in making the decision. The supervisors and the production manager are all white. As for the PDP, final selection of participants is made by incumbent foremen and superintendents. There are no formal, written guidelines for the selection procedure.
In Rowe v. General Motors Corp., 457 F.2d at 358-59, this Court concluded that an employer’s promotion and transfer procedures violated Title VII. We held that five aspects of the selection process, when taken together, had a discriminatory effect on black employees: (a) the foremen’s recommendation was the single most important factor in the promotion process; (b) foremen were given no written instructions pertaining to the qualifications necessary for promotion; (c) controlling standards were vague and subjective; (d) hourly employees were not notified of promotion opportunities or of the qualifications necessary to get jobs; and (e) there were no safeguards in the procedure designed to avert discriminatory practices.
Stockham’s system for selecting supervisors is basically the same as the system condemned in Rowe. First, incumbent foremen and superintendents play a critical role in the selection of new members of the supervisory staff and PDP participants. This role is even greater than the one criticized in Watkins where we also applied the five Rowe factors. See Watkins v. Scott Paper Co., 530 F.2d at 1193. Second, the recommendations of incumbent supervisors are largely discretionary and there are no adequate safeguards against racial bias. Third, there are not written guidelines specifying the necessary qualifications of a supervisor for either the direct selection process or the PDP. Fourth, as found in Watkins, some of Stockham’s criteria for promotion can be described only as subjective. Certainly the primary one, “best man”, has no objective elements. Fifth, there is no systematic procedure at Stockham for informing employees of supervisory vacancies. There has never been a system of job posting for any kind of vacancies at Stock-ham despite union demands that the procedure be implemented. The district court apparently considered the timely application procedure an adequate alternative. We do not. First, the “useless act” or “meaningless request” phenomenon discussed above54 impedes the procedure’s effectiveness. See Teamsters, 431 U.S. at 368, 97 S.Ct. 1843; Bing v. Roadway Express, Inc., 485 F.2d at 451. Moreover, of 75 employees selected as foremen since 1965, only two filed timely applications. The company itself does not take the procedure seriously.
The district court excused the grossly disproportionate number of black supervisors on the notion that there has been a low turnover in such positions since 1965. 394 F.Supp. at 478, 496. On the contrary, the evidence shows that over 63 percent of the present foremen were selected after 1965. We have concluded that Stockham’s supervisory selection procedure is discriminatory, in part, because of the grossly dispropor*347tionate number of black supervisors and the similarities between the discretionary, subjective selection procedure at Stockham and the system condemned in Rowe. The district court’s contrary findings are clearly erroneous.
The plaintiffs also challenge the process by which participants in the management training program (“MTP”), and its predecessor organizational apprentice program (“OAP”), are selected. These programs were designed to train college graduates in technical fields for upper level management positions. The defendant recruits participants from largely all-white universities such as Auburn University, the University of Alabama, the University of Tennessee, Georgia Institute of Technology and Sam-ford University. Stockham has never recruited at the six or more predominantly black institutions in the area. No black has ever been selected for this program although fifty whites have been chosen since 1969.
In United States v. Georgia Power Co., 5 Cir. 1973, 474 F.2d 906 at 926, this Court condemned the concentration of a company’s recruitment activities at white educational institutions:
The company’s policy of seeking skilled personnel only at white educational institutions is similarly an invidious brake on black employment opportunities for which no business necessity justification was shown. While the company obviously ought not be enjoined to recruit on all college campuses unless it chooses to do so, it also ought not be allowed to continue to restrict its recruitment programs to all — or preponderantly all — white institutions while maintaining such a racially imbalanced work force.
We have reached the same conclusions here in deciding that Stockham’s recruitment system for the MTP and its predecessor, the OAP, operates as a “built-in-headwind” against blacks. Id. at 925.
F. Seniority System
The plaintiffs and the EEOC contend that Stockham’s departmental seniority system, when combined with the company’s discriminatory job assignment practices, unlawfully reinforces and perpetuates prior discrimination. According to the EEOC, the seniority system operates to “freeze in” the effects of discriminatory job assignments in two ways:
(1) by conditioning transfer on the sacrifice of seniority, it inevitably inhibits black employees from seeking to escape from departments to which they were assigned because of race, and
(2) by refusing to recognize after transfer seniority acquired in black departments, or departments with particular jobs open to blacks, it continues to penalize them for their initial, racially-based assignments.
We discussed Stockham’s job assignment practices in subsection “III.B.” of this opinion and concluded that the company has unlawfully allocated jobs on the basis of race both before and after the effective date of Title VII, at least until 1973. The issue here is whether Stockham’s departmental seniority system, which is neutral on its face, is unlawful because it accentuates the effects of Stockham’s discriminatory job assignment practices.
The seniority system at Stockham has been modified twice since 1970. Until June 10, 1970, an employee who voluntarily transferred to a new department lost all of his accumulated seniority at the time of the transfer. He was treated for promotion and regression purposes as a new employee. With the 1970 modification to the system, an employee transferring to a new department was given eighteen months from the date of the transfer to decide whether he wanted to return to his old department. If he decided to return, the employee was permitted to re-enter his old department with his accumulated seniority within twenty-four months of his transfer. If he stayed in the new department, the employee lost the seniority he had accumulated in his old department. In 1973 the collective bargaining agreement effected a further modification of the seniority system at Stockham. If after eighteen months an *348employee elected to remain in the department to which he transferred, he was permitted to retain his seniority in the old department for layoff purposes, but only until he had been in the new department as long as he had been in the old department. If during that period the employee was laid-off from the new department, he was allowed to return to his old department with his accumulated seniority. Even after the 1970 and 1973 modifications to the seniority system several features remained unchanged: (1) at some point after an employee transfers to a new department, he forfeits his accumulated seniority; (2) an employee who transfers between departments is a new employee for all promotion and regression purposes in his new department; and (3) a departmental employee has the first opportunity to promote to all vacancies within his department.
Seniority systems such as the one at Stockham consistently have been condemned by the courts because black employees must choose to commit “seniority suicide” to enter departments from which they were previously excluded unlawfully on account of race. In addition, the plaintiffs argue, the Stockham system also “locks” such employees into their old departments by not providing for pay rate retention, thereby forcing transferring employees to take a short-term pay cut to receive potentially greater earnings in the future.
According to the district court, “the record does not support the conclusion that Stockham’s departmental seniority system locks black employees into particular jobs, pay categories or departments.” 394 F.Supp. at 454. The court’s determination was based on two sets of findings. On the issue of the comparative desirability of predominantly black and predominantly white departments, the court found:
The relatively high earning opportunities in the foundries (malleable, brass, and grey iron) suggest that the high number of blacks in these departments results from voluntary choices by these employees to work in those departments where the most money can be made. . . . The evidence introduced by plaintiffs failed to establish that there are demonstrably superior working conditions in certain departments, and this Court finds that the working conditions in various departments at Stockham are generally the same.
394 F.Supp. at 453-54. As to the number of interdepartmental transfers made by black and white employees the court found:
At all relevant times black employees have accounted for the large majority of all inter-departmental transfers. Black employees accounted for a low of 59.4% in 1965 and a high of 89.7% in 1968 of all inter-departmental transfers.
394 F.Supp. at 452.
The district court’s conclusion that blacks have no incentive to transfer at Stockham because the departments to which they have been assigned are no less desirable than those with predominantly white employees is clearly erroneous. We have already discussed55 the" evidence on the working conditions of predominantly black and white departments. The record shows that black employees at Stockham tend to work in the hottest, dustiest, and dirtiest departments. In addition, the vast majority of blacks work in lower classified jobs under the company’s incentive system. Even if some black employees earn as much as whites working in higher job classes, the evidence shows they do so only under the great pressure of the “racetrack” while performing numbingly repetitive tasks. Further, the evidence does not establish that the earnings opportunities are the same in these departments. Thirty-three percent or 174 of all whites work in the seven departments with the highest earnings opportunities while only 4 percent of the blacks are employed in those departments. And, 70 percent or 836 of all black employees work in the seven departments with the lowest earnings opportunities; whereas, 23 percent of all white workers are employed in those *349departments. Therefore, the district court’s first set of findings does not support its conclusion that blacks remain in the departments to which they are assigned because they have no incentive to leave.
The statistics cited by the court on the percentage of interdepartmental transfers by black employees do not include the actual number of interdepartmental transfers. It is impossible, therefore, to judge the magnitude of the transfers at Stockham. Because some blacks are willing to transfer to new departments even though they will lose seniority and suffer a cut in pay as a result does not mean that many more blacks are deterred from transferring because of these disadvantages. In addition, we would expect a larger percentage of blacks than whites to transfer where, as in this case, whites already work in jobs with better earnings potential, less pressure, and more satisfactory working conditions; and the statistics cited by the district court support this thesis. Finally, the figures cited by the district court are not accompanied by information on the seniority of transferring blacks. The more junior the employee, the less inhibition to transfer there is from the system. The evidence cited by the district court, therefore, does not support its conclusion that the seniority system at Stockham does not “lock” blacks into jobs discriminatorily assigned to them.
The district court’s conclusion that the seniority system at Stockham had no “locking-in effect” because there was no incentive for black employees to transfer to new departments and because a higher percentage of blacks than whites changed departments after 1965 is clearly erroneous. In Pettway v. American Cast Iron Pipe Co., 494 F.2d at 223-224, this Court evaluated a departmental seniority system. In that case a transferring employee retained his former seniority for purposes of returning to his old department in ease of lay-off from his new department, but did not carry over any of his accumulated seniority for purposes of departmental promotion. Under that system an employee who transferred to a new department to enhance his chances for eventual advancement, higher pay, or better working conditions was required to endure a loss of seniority and a wage cut as a condition of transfer. We concluded in Pettway that the present effect of such a system was to lock black employees into jobs and departments to which they had originally been assigned on account of race. In so doing this Court rejected the district court’s eight reasons for finding that blacks were not being “locked-in”. Those reasons included voluntary refusal of blacks to accept training and promotion opportunities, their failure to request promotions, poor job performances by black employees, lack of job vacancies, and the lack of motivation on the part of blacks at the company. This Court determined that critical examination of those variables led to the conclusion that they did not weigh heavily enough to overcome the plaintiffs’ empirical evidence of racial stratification in departments, jobs, and pay rates.
Here too we reject the reasons given by the district court for finding no locking-in effect. The continued stratificaton of blacks in the least desirable departments and jobs, in terms of working conditions and earnings potential, is clear from the plaintiffs’ statistics.56 “Once it has been determined that blacks have been discriminatorily assigned to a particular department within a plant, departmental seniority cannot be utilized to freeze those black employees into a discriminatory caste.” (Footnote omitted.) Johnson v. Goodyear Tire & Rubber Co., 491 F.2d at 1373. Under the seniority system in effect at Stockham before 1970 blacks who transferred between departments lost all accumulated seniority for promotion and regression purposes. Blacks transferring to new departments after 1970 have been required to start as new employees for job promotion and regression in the new department. After a period of time in the new department a new employee is forced to forfeit all his accumulated seniority even for layoff back to his former department.
*350In United States v. Jacksonville Terminal Co,, 451 F.2d at 453, we dealt with a craft and class seniority system in which an employer who gained a position in a new craft or class could not retain his accumulated seniority in his former craft or class. In finding the system unlawful we commented: “In any industry loss of seniority is a critical inhibition to transfer.” That statement applies here too. Further, from 1965 to 1971 the company required a lower score on the Wonderlic test for an employee transferring to a job within his department than for an employee transferring from another department. This procedure operated more decisively to freeze in past discrimination than the three-day preferential bidding period for departmental employees condemned in Pettway. See Pettway v. American Cast Iron Pipe Co., 494 F.2d at 240. Finally, Stockham’s seniority system does not provide for rate retention for an employee who transfers to a lower paying job in a new department as a first step to reaping long range job and earnings rewards. These factors compel us to conclude that the departmental seniority system at Stockham, even as modified in 1970 and 1973, locks-in the effects of pre-and post-Act discriminatory employment practices at Stockham. See Johnson v. Goodyear Tire & Rubber Co., 491 F.2d at 1373-74 and cases cited at 1373 n.27; United States v. Georgia Power Co., 5 Cir. 1973, 474 F.2d 906, 926-27; United States v. Jacksonville Terminal Co., 451 F.2d at 452-453.
The defendant Stockham contends that even if the seniority system operates to lock-in the effects of past discrimination, the system is justified by business necessity. Although this contention is not detailed in the defendant’s brief, its essentials seem to be that the seniority system is a necessary element of the “unique multi-plant complex” operated by Stockham in Birmingham. A company official testified that the six “plants” are composed of the following seniority units: cast iron fittings, malleable fittings, bronze valve, iron valve, steel valve, and butterfly valve. He asserted that interdepartmental, or “cross-plant”, transfers will create substantial difficulties for Stockham.
The business necessity defense, as the defendant applies it here, has no substance. First, at least four of the twenty-two seniority departments at Stockham extend over more than one “plant”. The valve machining and assembly department extends over several of the six “plants” and the electrical, machine shop, and construction departments extend over all of the “plants”. Thus, the functional separation of the company’s manufacturing operations does not reach the proportions suggested by Stockham’s contention that it effectively operates six different plants at its Birmingham facility. Second, Stockham’s statement that it has not discouraged interdepartmental transfers is inconsistent with its assertion that there are difficulties associated with so-called “cross-plant” transfers.
To be justifiable under the business necessity, doctrine a seniority system must be essential to the goals of safety and efficiency. United States v. Bethlehem Steel Corp., 446 F.2d at 662; Pettway v. American Cast Iron Pipe Co., 494 F.2d at 245. The defendant has failed to satisfy that heavy burden. There has been no showing that departmental seniority is essential to safety and efficiency at Stockham.
In its recent opinion in Teamsters, 431 U.S. at 349, 97 S.Ct. 1843, the Supreme Court stated that a seniority system, not justified by business necessity, that operates to freeze the status quo of prior discriminatory employment practices would seem to be unlawful under the rationale of Griggs v. Duke Power Co. Nevertheless, the Court concluded that such a system is legally valid under section 703(h) of Title VII, 42 U.S.C..§ 2000e-2(h), if it is a bona fide seniority system within the meaning of that section.57 In reaching this result the *351Court refused to distinguish between seniority systems that perpetuate pre-and post-Act discrimination. Id. at 348 n.30, 97 S.Ct. 1843. As to relief, however, the Court differentiated between pre- and post-Act discriminatory acts, pointing to the holding in Franks v. Bowman Transportation Co., 1976, 424 U.S. 747, 96 S.Ct. 1251, 47 L.Ed.2d 444, that:
§ 703(h) does not bar the award of retroactive seniority to job applicants who seek relief from an employer’s post-Act hiring discrimination.
Id. at 346, 97 S.Ct. at 1860. Thus, according to the Court,
[p]ost-Act diseriminatees . . . may obtain full “make whole” relief, including retroactive seniority under Franks v. Bowman, supra, without attacking the legality of the seniority system as applied to them.
Id. at 347, 97 S.Ct. at 1860. See EEOC; Interpretative Memorandum, 7/12/77, 46 L.W. 2028.
In this case the plaintiffs have proved that Stockham engaged in post-Act discriminatory employment practices in job allocation, testing, craft training and selection, and promotion. Any black injured by these practices is entitled to relief, including retroactive seniority. The issue whether the seniority system at Stockham is bona fide under section 703(h) of Title VII is relevant to this case only if the district court concludes that the class of black employees represented by the plaintiffs consists of some blacks who suffered only from preAct discrimination. The following comments are intended to guide the district court’s analysis of that issue.
In Teamsters the Court focused on whether a seniority system, contained in collective bargaining agreements between the employer, a nationwide common carrier of motor freight, and the unions was bona fide. Under that system an employee’s seniority for competitive purposes was his bargaining unit seniority, which controlled the order in which he was laid-off or recalled and the order in which he could bid for a particular job. Line drivers were in one bargaining unit and city drivers and servicemen were in another. The practical effect of the seniority arrangement was that a city driver or serviceman who transferred to a line-driver job forfeited all the competitive seniority he had accumulated in his previous bargaining unit and started at the bottom of the line-driver’s roster. The Court found that the system locked-in the effects of the employer’s past intentional discrimination; nevertheless, the system was bona fide. The Court explained:
[The seniority system] applies equally to all races and ethnic groups. To the extent that it “locks” employees into non-line-driver jobs, it does so for all. The city drivers and servicemen who are discouraged from transferring to line-driver jobs are not all Negroes or Spanish-surnamed Americans; to the contrary, the overwhelming majority are white. The placing of line drivers in a separate bargaining unit from other employees is rational, in accord with the industry practice, and consistent with NLRB precedents. It is conceded that the seniority system did not have its genesis in racial discrimination, and that it was negotiated and has been maintained free from any illegal purpose. In these circumstances, the single fact that the system extends no retroactive seniority to pre-Act discriminatees does not make it unlawful. (Footnote omitted.)
International Brotherhood of Teamsters v. United States, 431 U.S. at 355, 97 S.Ct. at 1865.
As we read the Teamsters opinion, the issue whether there has been purposeful discrimination in connection with the establishment or continuation of a seniority system is integral to a determination that the system is or is not bona fide. See also *352United Air Lines, Inc. v. Evans, 431 U.S. 553, 97 S.Ct. 1885, 52 L.Ed.2d 571 (1977). The Court’s analysis suggests that the totality of the circumstances in the development and maintenance of the system is relevant to examining that issue. See also Washington v. Davis, 426 U.S. 229, 96 S.Ct. 2040, 48 L.Ed.2d at 608-09. In Teamsters the Court focused on four factors:
1) whether the seniority system operates to discourage all employees equally from transferring between seniority units;
2) whether the seniority units are in the same or separate bargaining units (if the latter, whether that structure is rational and in conformance with industry practice);
3) whether the seniority system had its genesis in racial discrimination; and
4) whether the system was negotiated and has been maintained free from any illegal purpose.
The Court analyzed the context in which the seniority system developed. In discussing the relationship of seniority units to bargaining units, the Court quoted a National Labor Relations Board opinion that emphasized the rationality of separate bargaining units in the case of over-the-road and city drivers “where they are shown to be clearly defined, homogeneous, and functionally distinct groups with separate interests”. Id., 431 U.S. at 356, n.42, 97 S.Ct. at 1865. Thus, the facts of a particular seniority unit are critical to a determination whether the system is bona fide; and a case-by-case analysis of seniority systems in light of section 703(h) is necessary.
In Teamsters the plaintiffs conceded that the seniority system did not have its genesis in racial discrimination and that it was negotiated and maintained free from any illegal purpose. There is no such concession here. The seniority system at Stockham was adopted in a collective bargaining agreement in 1949, when segregation in the South was standing operating procedure. The history of the negotiations associated with Stockham’s seniority system is clouded. Since 1967 the local union has sought major revisions in the departmental seniority system through contract negotiations with the defendant Stockham.58 In 1970 *353the union struck for five months seeking the company’s agreement with its proposals, including plant-wide seniority. Stock-ham’s failure to go along with revisions in the seniority system must be evaluated in the context of the company’s extensive unlawful employment practices during the period of the negotiations and its intransigent adherence to wide-spread segregated facilities at the plant, at least until 1974. In addition, Stockham’s resistance to revisions in the seniority system must be considered in light of the union's firm support for such changes and its willingness to strike for the proposed modifications in 1970. Finally, unlike the facts of Teamsters, here the seniority units do not reflect existing separate and distinct bargaining units that conform to industry practice. The district court should give careful consideration to the negotiations involving the seniority system at Stockham and to the employment practices of the company underlying such negotiations. Because the issue whether the seniority system is bona fide was not legally relevant when this case was tried, on remand the district court should allow the parties to present additional evidence on the question.
IV.
UNION LIABILITY
As to the defendant unions, Local 3036 and the United States Steelworkers of America, AFL-CIO, the district court concluded:
Since this Court finds that there has been no violation of the Act, there is of course no liability on the part of the defendant unions. However, in the interest of judicial economy, this Court concludes that if liability existed, and since the contract between Stockham and the unions is the product of negotiation, the defendant unions would also be liable in their roles as bargaining agent in the collective bargaining process.
394 F.Supp. at 499. We remand the question of the unions’ liability to the district court. The court clearly erred in finding no liability for any defendant under Title VII. Thus, the determination that the unions are not liable because the employer is not liable cannot stand. The alternative finding that the unions violated Title VII if the employer did also must fall. The Supreme Court’s decision in Teamsters may require a determination whether the seniority system at Stockham is bona fide. If the system is found to be protected under section 703(h) or if the court does not reach that question, the unions could not have violated Title VII in agreeing to and maintaining the system. Teamsters, 431 U.S. at 356, 97 S.Ct. 1843. If, on the other hand, the district court concludes that the seniority system is not bona fide, it must consider the question whether the unions’ role in ratifying collective bargaining agreements containing the various seniority provisions compels a finding of liability.
The plaintiffs in this case assert that the findings of the district court on the unions’ role in negotiating the seniority system are not supported by the evidence. The court found:
While other forms of seniority may have been discussed from time to time, the basic (i. e., written) bargaining proposals submitted to Stockham by the Unions have never requested a change from the existing departmental seniority system and have instead reflected an intention on the part of the unions to preserve the present seniority system.
*354394 F.Supp. at 451. The plaintiffs contend that the record reveals that from 1967 to 1973 the unions sought major revisions in the departmental seniority system through negotiations.59 Union segregation is not an issue in this case, as it was recently in Myers v. Gilman Paper Corp., 5 Cir. 1977, 544 F.2d 837. Since World War II a majority of the local union’s members have been black, and a majority of the members of the grievance committee and the officers have been black since 1967. On remand, the district court should evaluate these facts in light of Teamsters and Evans in determining whether the unions must bear legal responsibility in the event the seniority system is not bona fide.
V.
RELIEF
Congress has granted courts plenary equitable powers under Title VII, 42 U.S.C.A. § 2000e-5(g), and section 1981 for constructing appropriate remedies for employment discrimination. Pettway v. American Cast Iron Pipe Co., 494 F.2d at 243. In Jacksonville Terminal we emphasized that the responsibility for fashioning effective relief lies with the district court:
Under the District Court’s careful supervision, the parties should work together in developing effective antidotes for the racial discrimination which we perceive . . The District Court bears the ultimate responsibility for fashioning such relief; it is not limited to simply parroting the Act’s prohibitions but is required to order such affirmative action as may be appropriate.
United States v. Jacksonville Terminal Co., 451 F.2d at 458. See also Gamble v. Birmingham Southern R. R., 5 Cir. 1975, 514 F.2d 678, 685-686.
A. Injunctive and Declaratory Relief.
On remand the district court should grant the plaintiffs injunctive relief against present discriminatory policies and practices at Stockham. During the proceedings below the trial judge denied an injunction against the continued segregation of facilities at the Stockham. plant because he concluded the issue was “effectively resolved” by the conciliation agreement between the EEOC and Stockham. The district court’s finding that the women’s restrooms in the dispensary are not segregated is clearly erroneous, and an injunction must be entered against the continued segregation of the two restrooms.
The plaintiffs and the amicus EEOC contend that on the facts of this case broader injunctive relief is appropriate on the issue of segregated facilities. They argue that the psychological harm engendered by the involuntary separation of black and white people entitles the black employees at Stockham to the assurance of integration, which the injunction will provide. Further, the EEOC contends that the company's continued segregation of facilities in the ten. years following the enactment of the Civil Rights Act of 1964 suggests that Stockham did not make a good faith effort to comply with the Act. The Commission points out that the conciliation agreement does not cover all segregated facilities and only requires that the company make specified alterations in its plant as presently constructed. In addition, according to the EEOC the agreement does not comply with the Commission’s model draft requiring a statement from the employer that all facilities will be available to employees without regard to race.
This Court has held that absent clear and convincing proof of no reasonable probability of further noncompliance with the law a grant of injunctive relief is mandatory. Equal Employment Opportunity Commission v. Rogers Bros., 5 Cir. 1972, 470 F.2d 965. In Cypress v. Newport News General & Nonsectarian Hospital Ass’n, 4 Cir. 1967, 375 F.2d 648, 658, the Fourth Circuit made an observation, relevant here, that has been cited by this Court several times: “ ‘[Protestations or repentance and reform timed to anticipate or blunt the force of a lawsuit offer insufficient assurance’ that *355the practice sought to be enjoined will not be repeated.” See Rowe v. General Motors Corp., 457 F.2d at 359; Gamble v. Birmingham Southern R. R. Co., 514 F.2d at 683. In light of our reversal of its fundamental findings, the district court is directed to reassess the evidence relative to injunctive relief. Unless it can discern and identify clear and convincing evidence that there is no reasonable probability of further noncompliance, the district court should enter a broad injunction against the segregation of all facilities and programs at Stockham.
Injunctive relief against discriminatory selection and training practices and procedures for craft and supervisory positions is also necessary. The district court should enjoin the use of the high school education and age requirements for apprenticeship training until the requirements have been validated by Stockham. Watkins v. Scott Paper Co., 530 F.2d at 1182 and n. 31. The company must show that such screening devices are essential to the safety and efficiency of the plant. Pettway v. American Cast Iron Pipe Co., 494 F.2d at 245, citing United States v. Bethlehem Steel Corp., and at 250. See also Stevenson v. International Paper Co., 516 F.2d 103 at 116.
As we noted,60 employees are selected for the apprenticeship program by an overwhelmingly white supervisory staff applying such subjective criteria as “desire” and “aptitude”. In line with Jacksonville Terminal, we conclude that the company must ascertain and publicize objective qualifications for the apprenticeship training program. Watkins v. Scott Paper Co., 530 F.2d at 1194. In addition, the district court should direct Stockham to formulate from those qualifications written guidelines for use by supervisors in selecting candidates for apprentice training. Further, the court should mandate the development of procedures to assure that the selection practices are free of discrimination. See Rowe v. General Motors Corp., 457 F.2d at 359; Watkins v. Scott Paper Co., 530 F.2d at 1194.
Similarly, the district court should direct the development of objective, written guidelines for the selection of supervisors. These guidelines too should be publicized. Also, the court should advise the company to structure the selection process so that it includes safeguards to avert discrimination. See Rowe v. General Motors Corp., 457 F.2d at 359; Watkins v. Scott Paper Co., 530 F.2d at 1194. The court should also require the company to recruit management trainees at predominantly black colleges and universities within a reasonable distance from Birmingham.
As we have already observed, the timely application procedure is not an adequate replacement for a system of job posting. The district court should mandate the posting of all job vacancies and the qualifications for apprentice training and supervisory selection. See Rowe v. General Motors Corp., 457 F.2d at 360-61; Watkins v. Scott Paper Co., 530 F.2d at 1194.
Stockham terminated use of the Wonderlic Personnel Test and the Bennett Mechanical Comprehension Test in 1971. We leave to the district court the issue whether injunctive relief against such testing is appropriate now. See Albemarle Paper Co. v. Moody, 422 U.S. at 436, 95 S.Ct. 2362; Pettway v. American Cast Iron Pipe Co., 494 F.2d at 244. As for the Tabaka Tests, as stated, we remand the issue of the lawfulness of that testing program to the district court.
Seniority relief is mandatory in this case even if the district court finds the system bona fide. See Teamsters, 431 U.S. at 348, 97 S.Ct. 1843. The federal courts are empowered to fashion such retroactive seniority as the particular circumstances of a case may require to effect restitution and make whole, insofar as possible, the victims of racial discrimination in job assignments, transfers, and promotions. See Franks v. Bowman Transportation Co., 424 U.S. at 764, 96 S.Ct. 1251.
*356This Court has reiterated many times that “relief is to be granted within the framework of the ‘rightful place’ theory.” Gamble v. Birmingham Southern R. R. Co., 514 F.2d at 683. See also Local 189, United Papermakers and Paperworkers v. United States, 5 Cir. 1969, 416 F.2d 980, 988; United States v. Georgia Power Co., 474 F.2d at 927. In Franks v. Bowman Transportation Co., 424 U.S. at 760 n. 21, 96 S. Ct. 1251, the Supreme Court noted congressional approval of that theory in its enactment of the 1972 amendments to Title VII. The Court also recognized the “presumption in favor of rightful place seniority relief”. Id. at 779 n. 41, 96 S.Ct. at 1271. The “rightful place” doctrine requires that qualified blacks be given a preference for future vacancies in positions they would have occupied but for wrongful discrimination. United States v. Hayes International Corp., 456 F.2d at 118; United States v. Georgia Power Co., 474 F.2d at 927. We discussed the requirements of the theory in Bing v. Roadway Express, Inc., 5 Cir. 1973, 485 F.2d 441, 450, in the context of employment discrimination in job transfers:
Thus the rightful place theory dictates that we give the transferring discriminatee sufficient seniority carryover to permit the advancement he would have enjoyed, and to give him the protection against layoffs he would have had in the absence of discrimination. How much seniority the transferee deserves should be determined by the date he would have transferred but for his employer’s discrimination.
That date should be selected on the basis of when a black employee possessed the experience necessary to qualify for the job. Id. at 451. By showing a pattern of discriminatory practices against the class as a whole, the plaintiffs have established that there are reasonable grounds to infer that individual assignment, transfer, and promotion decisions were made pursuant to the discriminatory policy. As a result, Stock-ham is required to come forth with evidence to dispel the inference. See Teamsters, 431 U.S. at 358-365, 97 S.Ct. 1843; Baxter v. Savannah Sugar Refining Corp., 5 Cir. 1974, 495 F.2d 437, 443, cert. denied, 1975, 419 U.S. 1033, 95 S.Ct. 515, 42 L.Ed.2d 308. In Franks v. Bowman Transportation Co., the Supreme Court recognized the appropriateness of class-based seniority relief. The district court must conduct additional proceedings to determine the scope of individual relief. Teamsters, 431 U.S. at —, 97 S.Ct. 1843; Baxter v. Savannah Sugar Refining Corp., 495 F.2d at 444.
Where found to be appropriate by the district court, relief such as red circling61 and special training for discriminatees62 may be awarded. Other relief may include “an order that the employer keep records of its future employment decisions and file periodic reports with the court, or any other order ‘necessary to ensure the full enjoyment of the rights’ protected by Title VII.” (footnote omitted). Teamsters, 431 U.S. at —, 97 S.Ct. at 1867. Also the court may establish goals and set timetables in order to bring about Stockham’s earliest possible compliance with Title VII. See Morrow v. Crisler, 5 Cir. 1974, 491 F.2d 1053, 1056-57, cert. denied, 419 U.S. 895, 95 S.Ct. 173, 42 L.Ed.2d at 139.
B. Backpay.
Under 42 U.S.C. § 2000e-5(g) back pay may be awarded where appropriate to victims of unlawful employment practices. In this case the district court concluded:
The record in this case is barren of any persuasive evidence of tangible economic loss with respect to the class. Furthermore, plaintiffs have not proved the existence of any discriminatory condition or practice at Stockham which would effect earnings and therefore be relevant to the award vel non of back pay. Consequent*357ly, no back pay is awarded to plaintiffs and the class they represent.
394 F.Supp at 500. We have already determined that the district court erred in finding that the plaintiffs had failed to prove discrimination by Stockham. In addition, the court’s conclusion that there is no “persuasive evidence of economic loss” is also clearly erroneous. We therefore remand the issue whether the plaintiffs, and the members of the class they represent, are entitled to back pay to the district court. The court should be guided by the standards established by the Supreme Court in Albemarle Paper Co. v. Moody, 422 U.S. 405, 95 S.Ct. 2362, 45 L.Ed.2d 280, as well as by the decisions of this Court. In Moody the Court observed:
It is true that backpay is not an automatic or mandatory remedy; like all other remedies under the Act, it is one which the courts “may” invoke. . . . However, such discretionary choices are not left to a court’s “inclination, but to its judgment; and its judgment is to be guided by sound legal principles.” (footnote omitted).
422 U.S. at 415-16, 95 S.Ct. at 2370. According to the Court, a backpay award should be measured against the purposes of Title VII, to achieve equal employment opportunities:
It is the reasonably certain prospect of a backpay award that “provide[s] the spur or catalyst which causes employers and unions to self-examine and to self-evaluate their employment practices and to endeavor to eliminate, so far as possible, the last vestiges of an unfortunate and ignominious page in this country’s history.”
422 U.S. at 417-19, 95 S.Ct. at 2371. Therefore, when a finding of unlawful discrimination has been made,
backpay should be denied only for reasons which, if applied generally, would not frustrate the central statutory purposes of eradicating discrimination throughout the economy and making persons whole for injuries suffered through past discrimination. (Footnote omitted.)
422 U.S. at 421, 95 S.Ct. at 2373. As this Court explained in Carey v. Greyhound Bus Co., Inc., 5 Cir. 1974, 500 F.2d 1372, 1378, a district court’s discretion to deny backpay is narrow:
“Once a court has determined that a plaintiff or complaining class has sustained economic loss from a discriminatory employment practice, back pay should normally be awarded unless special circumstances are present.” (Footnote omitted.)
(quoting from Pettway v. American Cast Iron Pipe Co., 494 F.2d 252-53.) The “special circumstances” discussed by Judge Tuttle in Pettway are not applicable to this case. See Pettway v. American Cast Iron Pipe Co., 494 F.2d at 253-56.
A backpay inquiry proceeds in two phases.
During the first or general phase the class must show its entitlement to back-pay. This is presumptively shown once employment discrimination has been found. The second or specific phase allows each discriminatee to present the backpay claim to which he is presumptively entitled unless the defendant bears the burden of proving nonentitlement.
Swint v. Pullman-Standard, 539 F.2d at 103. For purposes of backpay relief economic harm is not a necessary element of the plaintiffs’ prima facie case “unless the defendant has shown ‘convincingly’ by ‘statistically fair exhibits’ that the class earned ‘at least as much as a plant-seniority comparable group of whites.’ ” Id. at 92. The defendant Stockham failed to meet this burden. The conclusion we reached in Johnson v. Goodyear Tire & Rubber Co., 491 F.2d at 1374-75, is equally applicable here and should guide the district court on remand:
Since we have held that all black employees hired into the labor department before April 22, 1971 were the victims of discrimination as a result of Goodyear’s testing, diploma requirements and transfer policies, we conclude as a matter of law that the members of this class of discriminatees are presumptively entitled *358to an appropriate award of back pay unless evidence is adduced to establish as a matter of fact that such discriminatory practices did not adversely affect a particular claimant (or claimants). On remand, the initial burden will be on the individual discriminatee to establish that he is a member of the class discriminated against and thus presumptively entitled to recover back wages under the facts pertinent to his individual claim. If an individual proves his claim and class status, the burden should appropriately shift to the employer to show by convincing evidence, extenuating circumstances, which would support the conclusion that the individual would never have transferred regardless of its employment practices. .
The plaintiffs seek “front pay” as part of the backpay award. In “front pay” relief, a monetary award is calculated to terminate on the date a victim of discrimination attains an opportunity to move to his “rightful place” rather than on the date the order granting relief is entered. See Patterson v. American Tobacco Co., 4 Cir. 1976, 535 F.2d 257, cert. denied, 429 U.S. 920, 97 S.Ct. 314, 50 L.Ed.2d 286; E. E. O. C. v. Enterprise Ass’n Steamfitters, Local No. 638, 2 Cir. 1976, 542 F.2d 579, app. pending. The First Circuit stated in E. E. O. C. v. Enterprise Ass’n Steamfitters, 542 F.2d at 590-91:
It is the date of actual remedying of discrimination, rather than the date of the district court’s order, which should govern. . . . We agree with the Government that to hold otherwise is to encourage the union to delay the remedial process rather than to encourage the rapid achievement of the discrimination victims’ rightful place.
We agree that there are cases in which such relief is appropriate. The district court should consider the suitability of awarding “front pay” relief in this case in light of the company’s long-term resistance to desegregation and to equal employment opportunity for blacks. This may well be one case in which “front pay” relief is factually compelled. In Patterson v. American Tobacco Co., 535 F.2d at 269, the Fourth Circuit noted that a backpay award can be supplemented with post-judgment monetary relief in several ways:
This compensation should be supplemented by an award equal to the estimated present value of lost earnings that are reasonably likely to occur between the date of judgment and the time when the employee can assume his new position. Alternatively, the court may exercise continuing jurisdiction over the case and make periodic back pay awards until the workers are promoted to the jobs their seniority and qualifications merit. Or perhaps counsel and the court can devise some other convenient method of taking all the effects of past discrimination into account.
If the district court concludes that “front pay” relief is required in this case, it should determine which of the Patterson approaches is most suitable to the facts of this case.
C. Attorneys’ Fees
The plaintiffs contend that this Court should direct the district court to make an interim award of attorneys’ fees because of the extensive nature of the litigation in this case. We agree. This suit was initiated on October 5, 1966; the trial occurred in February 1974; this Court has decided the liability issues in this opinion. We have emphasized the key role played by “private Attorneys General” in actions involving employment discrimination. See Rowe v. General Motors Corp., 457 F.2d at 360 n. 26; Gamble v. Birmingham Southern R. R. Co., 514 F.2d at 686. In addition, the Supreme Court has stressed the “strong public interest in having injunctive actions brought under Title VII, to eradicate discriminatory employment practices.” Albemarle Paper Co. v. Moody, 422 U.S. at 415, 95 S.Ct. at 2370. There is a danger that litigants will be discouraged from bringing such suits because of the risks of protracted litigation and the extended financial drain represented by such a risk. An award of interim attorneys’ fees will prevent ex*359treme cash-flow problems for plaintiffs and their attorneys. This Court has recognized that interim awards are proper in appropriate cases. Baxter v. Savannah Sugar Corp., 495 F.2d at 447. Here, where the litigation has consumed more than eleven years such an award is appropriate. Otherwise, the danger exists that defendants in Title VII suits may be tempted to seek victory through an economic war of attrition against the plaintiffs.63
We reverse the district court’s finding that the plaintiffs are entitled to attorneys’ fees only in an amount proportionate to the assistance their suit provided to the achievement of a conciliation agreement between Stockham and the EEOC. We remand the issue of attorneys’ fees to the district court with instructions to award interim fees to the plaintiffs covering all pre-trial, trial, and appellate work by their attorneys.64 This amount will of course be subtracted from the sum finally awarded for attorneys’ fees for the full course of the litigation.
VI.
CONCLUSION
The record in this ease reveals that the defendant Stockham is guilty of unlawful racial discrimination in the segregation of facilities and programs, the allocation of jobs, craft training and selection, promotion, and supervisory recruitment and training. We have discussed the broad remedies available to the district court under Title VII for rectifying such practices. Except for backpay relief, nothing the courts can do will change the past. On remand, the future of blacks at Stockham is at issue.
REVERSED AND REMANDED.