PRELIMINARY STATEMENT
This cause was commenced on March 20, 1975, in the United States District Court for the District of New Mexico, on which date the United States Department of Labor filed a complaint against Sandia alleging age discrimination under Section 7(b) of the Age Discrimination in Employment Act, 29 U.S.C. Section 621, et seq. Also alleged to be violative of the ADEA was an administrative practice maintained by Sandia until April 30, 1973, which mandated intervals between salary increases and known as “stretch-out.” Subsequently, on September 16, 1975, this action was consolidated with two private actions, previously pending in the trial court, which actions arose from the same occurrence, namely, a work force reduction conducted by Sandia which had commenced in the spring of 1973.
This action is to be distinguished from these previously pending private actions which are also on appeal in this court in our Nos. 79-1021, 79-1168 through 79-1181, Mistretta, et al. v. Sandia Corporation, which were consolidated with this case by order of this court on March 9, 1979, and argued together with this case. This case is the principal action, and the private actions are governed by the law and evidence of this action, although additional legal and factual issues are presented by the private actions which are not considered here.
The three consolidated actions proceeded as a “spurious class action” under the provisions of 29 U.S.C. Section 216 rather than as a class action under Rule 23 F.R.C.P. Because of the multiplicity of parties and issues involved in the continuous filing of claims in the government’s action and of consents to become plaintiffs in the private actions, possibly totaling 225, the issues of individual claims and damages and whether claims were timely filed were postponed, and the Secretary of Labor’s action to establish a pattern or practice of unlawful discrimination, along with the same issue from the actions of the private plaintiffs, was tried to the court, Edwin L. Mechem, Judge, on January 19 through February 17, 1977. The object was to divide the “liability” and “remedial” issues involved in the various claims. The sole question to be determined was whether a prima facie case of age discrimination in Sandia’s employment practices could be shown.
On October 20,1977, the trial court issued an Interlocutory Opinion in which it held, inter alia, that a prima facie case had been established showing that Sandia had engaged in a pattern or practice of discrimination against a portion of the protected class, those employees between the ages of 52 and 64. The opinion directed that at the subsequent trial, Sandia would proceed and have the burden of presenting evidence in its defense, responding to the prima facie case and overcoming it. Having made findings and conclusions, partially based upon the statistical evidence, the trial court held that
It had been previously decided that individual claims in the private action would follow the trial on pattern or practice issues. Individuals age 40 to 51 who filed consents in the private action may present their individual cases that age was a factor in their selection for layoff. They, however, have the burden of putting on a prima facie case of discrimination and must carry the burden of proof. Consent plaintiffs laid off between March 23, 1973 and October 18, 1973, in the 40-51 age group, must also prove that the notice requirement should not bar their complaints. These burdens must also be carried by individuals ages 40-65 who filed consents and who were laid off involuntarily prior to the March, 1973 reduction in force.
This concludes the liability phase of this action. In the remedial phase, the focus will be on individual relief. Sandia may present evidence showing that each individual was selected for layoff based on reasons other than age, subject to further evidence by the Government or consent plaintiff showing that reasons are pretexts for unlawful discrimination. International Brotherhood of Teamsters v. United States, 431 U.S. 324, 97 S.Ct. 1843, 52 L.Ed.2d 396 (1977).
*603Pursuant to the above Interlocutory Order, the public and private suits were separated for trial in the “second phase.” Trial of the issues involved in the actions brought by the first seventeen private plaintiffs commenced on June 20, 1978 and continued through July 6, 1978. Trial of the issues raised by the second group of fourteen plaintiffs claiming through the Secretary of Labor was held from October 24 through November 2, 1978. Subsequent trials were held thereafter.1 On May 31, 1979, a “partial final judgment” was entered on findings and conclusions entered February 20, 1979, in favor of the Secretary of Labor on the claims of eleven former Sandia employees in the total sum of $333,614 for lost earnings, $9,665 for lost savings contributions, and other relief, which judgment forms the basis for this appeal. On July 1, 1979, the United States Equal Employment Opportunity Commission succeeded to the enforcement authority of the Secretary of Labor, 3 C.F.R. 321 (1979), which fact, it is agreed by the parties, has no effect upon the issues which are to be considered here. Subsequently, the EEOC succeeded the Secretary of Labor as a party herein. The judgment considered here granted injunctive relief, back pay, certain benefits, and reinstatement to eleven employees, all of whom were between the ages of 52 and 64 and had been terminated by Sandia in 1973. Issues relating to damages on behalf of present and former Sandia employees arising as a result of certain salary administration practices known as “stretch-out,” are still pending in the trial court. As to the claims of two claimants, the trial court found in favor of Sandia and the claims were dismissed. The record is incomplete as to the disposition of the claim of one claimant, Donald E. Fossum.
SUMMARY OF FACTS
The factual background which has given rise to this case is essential to an evaluation of the issues raised in this appeal. Sandia was first organized in 1946 as an engineering and testing division of Los Alamos Scientific Laboratories which were then engaged in the field of developing atomic weaponry. In 1949, it was incorporated as a wholly owned subsidiary of Western Electric Company to serve as prime contractor engaged in engineering and testing for the United States Atomic Energy Commission. All of Sandia’s contracts and work are federal and all of its funding, including funding for compensation of personnel, derives from federal appropriations. During the 1960’s, Sandia engaged in research and development activities as assigned. Now it is engaged in the development of new technologies, including nuclear weaponry, nuclear power and safety, and solar energy technology and projects. Its primary responsibility continues in the area of nuclear ordnance.
The Atomic Energy Commission has now become the United States Energy Research and Development Administration, and the functions of Sandia are under the jurisdiction of the United States Department of Energy. By the terms of its contract with ERDA, Western Electric does not receive profits from Sandia’s operations.
Inasmuch as Sandia’s sole funding is federal, it undergoes change as a result of federal policies and political needs. In 1968, Congress instituted cutbacks in project funding in activities of Sandia. A work force reduction was required to be instituted at Sandia in 1970, with existing employees being terminated both voluntarily and involuntarily. In 1971, another program was instituted to terminate some employees involuntarily, and this program was examined but left unchallenged by the Department of Labor. It was specifically approved by the EEC compliance section of the AEC.
*604In 1972, however, a further budget cut for the fiscal year 1974 indicated to a committee of administrators formed by Sandia that its 12% reduction in staff which had been achieved was going to be insufficient: approximately 800 additional employees were to be terminated. On January 29, 1973, the threatened reduction which led to this action took place. Management announced to the employees that a further force reduction program was to be instituted during the spring of 1973.
The resulting program was widespread and painful. Each of Sandia’s vice presidents was assigned a quota of existing personnel to nominate for termination, and supervisors were required to review and evaluate existing employees. Only staff or technical employees and the “associate” and “assistant” staff and technical level employees (designated as “non-graded” employees) were involuntarily terminated. Prior to the reduction in force there were approximately 4500 “non-graded” employees. The program terminated 810 employees; of these, 320 were terminated involuntarily; 306 of these were classified as “non-graded.” A total of 207 of the 306 terminated “non-graded” employees were age 40 or older.
Of the employees who were terminated voluntarily, some left by attrition and others were permitted to take advantage of enhanced pension and retirement benefits or layoff allowances. Some were granted a reprieve to reach pension vesting anniversaries before March 1974. The fates of 39 were spared, 38 of whom were in the protected age group.
The procedures followed in selecting existing employees for involuntary termination were substantially the same as those followed in the previous force reduction plan which had gone unchallenged by the Labor Department. In the early stages the procedure was for supervisors at various levels to nominate “candidates” for termination. The supervisors themselves were not, of course, subject to the reduction process. Nevertheless, some supervisors were required to accept demotion to non-supervisory jobs without a change in salary.
The names of candidates nominated by the supervisors were reviewed by two specially formed committees consisting of personnel selected on a company-wide basis. Instructions to the committee members were to submit performance evaluations for each candidate. Sandia claimed that some employees were terminated because their jobs were eliminated. The trial court doubted whether any of these were terminated for that reason since the duties of those jobs were merely reassigned.
Because of previous complaints as well as humanitarian considerations, the ages of employees nominated for termination were considered by the review committees. The Age Discrimination in Employment Act was not overlooked as a problem. The committees reviewed the candidates in light of personnel records and the supervisor’s performance evaluations, as well as whether detailed layoff criteria were being employed by the various departments within Sandia.
Final authority for the layoff resided in a committee of vice presidents of Sandia, called the “Small Staff,” which reviewed the candidates, compiled a “master list” of candidates for termination, and issued the layoff notices, effective March 23,1973. In some cases, the effective dates were on benefit anniversary dates. The present action and its consolidated cases, as well as those cases and issues in this case still pending in the district court, concerns the results and ramifications of the actions.
SCOPE AND FORM OF FINDINGS OF FACT
On February 20, 1979, the trial court entered its Findings of Fact and Conclusions of Law, to which minor modifications were later made, which are the basis of this appeal. The court noted therein that the issues relating to the alleged discriminatory impact of the “stretch-out” practice of salary administration were being reserved for a later trial and that only claims relating to involuntary termination of the employees represented by the Secretary of Labor were *605being adjudicated. The trial court both found and concluded that Sandia had had full knowledge of the applicability to its employment practices of the Age Discrimination in Employment Act from the time of its passage. Findings of fact were made as to the claims of 13 of the individuals. It concluded that Sandia had willfully violated the provisions of the ADEA in its discharge of 11 of the claimants, but had not violated the ADEA with respect to two others. The court’s further conclusion was that the Secretary of Labor was entitled to injunctive relief to prevent further violations of the ADEA and to restrain Sandia from withholding compensation found to be due to the claimants (denominated as “plaintiffs”), and that certain of the claimants were entitled to be offered reinstatement.
THE ISSUES
The three essential issues which are presented for review are:
1. Whether the trial court erred in finding that the evidence established a prima facie case that Sandia unlawfully discriminated against employees on the basis of age, and, specifically, whether the trial court erred in its consideration of the statistical evidence.
2. Whether the trial court erred in interpreting the effect of the finding of a prima facie case of a pattern and practice of illegal age discrimination as it applied to the burden which Sandia was required to bear in the “remedial stage” of the proceedings.
3. Whether the trial court erred in assessing and awarding relief to the claimants.
THE STATISTICAL EVIDENCE
The statistical evidence adduced at the “pattern or practice” trial was presented by plaintiffs through Dr. John M. Spalding, a statistician. Dr. Spalding had been retained to conduct statistical tests on the data relating to hiring practices, demotions and salary administration policies together with terminations which resulted from the reduction in force at Sandia.
According to Dr. Spalding, the field of statistics is comprised of two groups:
Descriptive statistics, which summarizes informational data by using averages and measures of dispersion.
Statistical inference, which uses the informational data to arrive at statements of probability.
His testimony tended to pursue the latter use of the summarized data. He used computer runoff sheets which contained employee data supplied by Sandia, making corrections to his original data. From this information, he prepared statistical bar graphs and OGIVE charts for each category of employee involved in the reduction in force, together with some combinations of categories, subcategories, and an overall study of all employees involved.2
The term “OGIVE” is not defined in the record, but the charts show the results of the application of the test, used in each of the analyses, known as the Kolmogorof-Smirnov test. That test was said to measure the difference between an observed frequency of the occurrence of an event and *606the frequency of the event which would be expected if the event had occurred by chance. The statistician then consults standard statistical tables and makes a statement as to probability of the event occurring solely by chance. This was also termed a test of the statistical significance of the “goodness of fit” of the occurrences. Statisticians for both plaintiffs and defendant agreed that the probability of a chance occurrence of less than 5% was statistically significant. Dr. Spalding conducted 36 separate analyses of the computer data with a view to establishing his “null hypothesis” that selection of an employee for inclusion in the layoff, all other factors being equal, was independent of age. The probability of the variable, age, being a chance occurrence was less than 5%, (in many cases less than 1%) in 34 of the analyses.3 This was “statistical inference” evidence which both sides employed to support their positions.
In addition to the computer runoff sheets, Dr. Spalding had access to, and used in his preparation of the graphs and charts, both a master list of employees who had been scheduled to be terminated and a confidential termination classification list which coded each employee involved in the reduction in force according to the method of disposition of his involvement. According to this confidential classification:
Employees who were terminated voluntarily were assigned a code of “A.”
Employees who left Sandia voluntarily, but who, unbeknownst to them, had been scheduled to be terminated involuntarily, were assigned code “B.”
Code “C” consisted of employees who had requested that their personnel records reflect a voluntary termination but who actually were terminated involuntarily.
Code “D” represented those employees who were involuntarily laid off and whose records reflected that fact.
A fifth category, added by the plaintiff, code “E,” consisted of those employees who were scheduled on the master list to be laid off but who, because more employees volunteered for layoff than had been anticipated, were “saved” or “snatched back” according to local parlance. All remaining employees were designated as code “F.”
Based upon the codes, the analysis proceeded according to the company designated job classifications. The employees here involved were considered to be at a level below both the supervisory and management level, but above the “graded” level assigned to clerical and manual personnel. These employees, considered to be “non-graded,” consisted of the scientists and engineers engaged in the research and development activities. These are designated Members of Technical Staff, or MTS. The technical staff employees were allowed to work with or without supervision, depending on their job performance ratings and other factors.
The remaining “non-graded” employees were designated “PEP plan” or Position Evaluation Plan Employees. For the most part, this class is made up of support or staff positions. PEP plan employees are classified as to whether their jobs are technical or administrative.
Technical employees are designated either Technical Staff Associates, TSA’s, who are professional but non-degree holding aides to the technical staff.
Staff Assistant Technical, SAT’s are technically trained assistants to the Technical *607Staff and Staff Associates, or engaged in testing, maintenance and production.
Administrative employees are designated either Member of Laboratory Staff, MLS, consisting of professional level administrators, low level supervisors, and professional support technicians concerned with activities outside the main missions of the company (these people may be supervised or non-supervised), or Staff Assistant Laboratory, SAL, who are assistants to the Laboratory Staff, drafters, designers, or senior clerical employees.
Within each of these job categories, and in two instances, subcategories based upon whether the employee was supervised or non-supervised, the codes were applied to the terminated and “saved” employees and the resulting data analyzed on the basis of the ages of employees on February 1, 1973, just prior to the layoff.
The tests were studies of the observed frequencies of the selection of employees eligible for termination under the announced reduction in force, considered by age, compared to the population of all employees within each job category. Three main combinations of termination codes were selected for study; C and D, which were employees who were involuntarily terminated; B, C and D, which were the employees who were involuntarily terminated and those who would have been so terminated had they not volunteered; and B, C, D and E, which were all of the employees included in the latter combination with the addition of the “saves” or “snatchbacks.” All three of these combinations concerned employees who were in fact selected by Sandia for termination under the reduction in force. In addition to these main combinations, other combinations, as well as each group of employees within a single termination code category were studied, but not presented in graphic form.
Separate analyses were conducted for the categories of employees, MTS plus all PEP employees, both with supervisors and without supervisors; all PEP employees, both with and without supervisors; MTS alone, both with and without supervisors; SAT; TSA; SAL; LSA; and MLS both with and without supervisors.4
The trial court based its consideration of the statistical evidence on a numerical and percentage overview of the effects of age on the data. It considered the strength of the expert testimony presented by the plaintiff to be in the “presentation of the data” rather than in terms of the probability statements and statistics. It gave considerable weight to the data relating to the C and D employees, those actually involuntarily terminated, and used the combined data which consisted of BCD and BCDE groupings, those selected for involuntary termination, only as corroborative. Its consideration of the data combined the analyses of the two job categories which had been separated on the basis of whether or not the employee was supervised or unsupervised. It commented generally on the data relating to each job category, consideration being given in some categories to the category’s size and uniformity of age range, the percentage of the category laid off in the reduction in force, the median age of the category’s employees and the median age of those terminated, the point at which age appears to have become a factor in the termination of the employees. In addition, the trial court considered documentary evidence of employee records which it regarded as significant and having an effect on the selection process. One element which the trial court considered in common in connection with each category was the age range within the protected group which it felt to be disproportionately affected by the terminations. This presumably subjective determination can best be demonstrated in the following columnar analysis:
*608Age range considered % of total terminated C&D % of nonulation within category
SAT
52- 64 53 20
TSA
53- 64 64 25
SAL
52-64 36 21
LSA
58-63 * 55 12
MLS (combined)
55-64 53 15
MTS (combined)
54- 64 11 3
Lab-wide study of PEP employees non-graded employees, all MTS and
54-64 33 15
In each of the instances in which the trial court considered the data introduced by the addition of the B (voluntary but would have been terminated anyway) category and E (selected for termination but “saved”) category, the disproportion was bolstered.
In addition to the consideration of statistical data, the trial court considered exhibits introduced by the plaintiffs, largely made up of management and supervisory memoranda and personnel reports, which tended to indicate that older non-supervisory employees were stereotyped as unproductive and as becoming technically obsolete; that the company’s management was convinced that “new blood” or young Ph.D.s were the company’s future due to the rapidity of technological changes. Also considered was “an arbitrary generalization” that older employees could not keep pace with new technology, an attitude which the trial court felt corroborated the statistics and supported the inference that age was a factor in the selection of employees for termination. A substantial number of layoff decisions were made by supervisors who were in the lower half of the protected age group. Sandia’s reviewing committees had noted the disproportionate impact of the layoffs on older employees, but were unsuccessful in solving the problem.
Plaintiffs supplemented their statistical evidence by calling a number of Sandia’s supervisors, who testified that management was concerned over the increasing average age of the corporation’s employees. The concern of Sandia over the increasing average age caused it to use various methods to encourage older employees to volunteer for the reduction in force program.
A division supervisor had a meeting, warned of the impending cutback and said that older people had better really consider retiring. At about this same time, another supervisor told one group of employees that “Any number of us ought to prepare to retire by age 55,” because he believed that it would be Sandia’s policy in a few more years.
Another witness said that one of his supervisors told him to remember that he had reached a dangerous age, telling him that he had reached 55 and “you can be forced to retire.”
Just prior to the commencement of the RIF program, the associate director, Pope, wrote a memo reporting how to encourage nonessential employees to leave the company. In the memo it was pointed out that people with 15 years’ tenure, between the ages of 48 and 55, even when given Sandia’s lowest performance ratings, are far above national averages and that their skills may well be valuable in many situations. This, continued Mr. Pope, was for encouraging such employees to leave Sandia.
There was also evidence that qualification for benefits under the retirement plan was a factor in choosing older employees for layoffs.
One department manager testified that he was told to consider three criteria in nominating employees for termination: elimination of jobs, job performance and the likelihood of personnel taking advantage of the early retirement program.
TESTIMONY ON BEHALF OF DEFENDANT SANDIA
In response to the testimony of the plaintiffs, Sandia offered expert testimony by its statistician, Dr. Prairie, whose theme was *609that the most significant factor in determining which of the employees were to be terminated was performance rather than age. He gave testimony based upon analysis of the same data which had been used by Dr. Spalding. It was carried out in four stages.
The first of these was a contingency analysis which considered the C and D groups (those involuntarily terminated) as compared with the A, B, E and F groups dividing the employees by job classification and by age only on the basis of whether the employees were within the protected class (over 40 or under 40). This step gave a statistically significant correlation between termination and age in the MLS and SAT job categories, but Dr. Prairie concluded that this fact only indicated that further analysis must be performed.
The second of the four steps was application of the Kolmogorov-Smirnov test to the MTS category based upon Dr. Prairie’s division of the employees into three subcategories; low performers, middle performers, and high performers, all based upon the defendant’s performance rating system. The trial court noted that this step would be considered flawed if the performance ratings contained age bias. Ultimately, the court rejected Dr. Prairie’s analysis on this basis. Dr. Prairie’s own finding was that employees under age 40 tended to receive higher performance ratings, but contended that much of this influence was due to a correlation between performance and formal education (Master’s degree level). Dr. Prairie’s opinion was that age was a factor in its impact on the termination of only those employees under age 40, and of those, only those classified as “middle performers.” 5
Dr. Prairie’s third step added an additional element which assigned a numerical value used by Sandia’s personnel system for education. This step, termed a “discriminant analysis,” sought to identify the factors or combinations of factors which contributed to employee selection for layoff. Individual age groupings were considered separately. This step convinced Dr. Prairie that performance was the key factor in determining which employees would be selected for layoff.
The fourth step divided the employee categories on the basis of whether the employees were over or under age 40. It ranked them according to performance levels assigned by Sandia; outstanding, good, satisfactory, and unsatisfactory. This step was, in addition, a discriminant analysis, done separately for the MLS, MTS, SAT, and SAL categories, and reaching the conclusion that there was a strong relationship between low performance and layoff. Here again, the significance between age and performance was partially explained by the influence of higher formal educational rankings being prevalent among younger employees.6 Dr. Prairie’s studies considered only employees actually terminated, and not those selected for termination. Sandia also submitted testimony based on a study designed to apply objective criteria to the performance rating system in order to show that the layoff patterns would have been similar had objective, rather than subjective, ratings been utilized.
The above is illustrative of the very extensive direct testimony on behalf of the plaintiffs which served to corroborate the statistics. The trial court, on October 20, 1977, entered a so-called Interlocutory Opinion which is the subject of objection and complaint on the part of Sandia. In it the trial court disposed of a number of pretrial problems which had not been finally ruled on.
THE TRIAL COURT’S FINDINGS
The trial court considered age discrimination as not as clear-cut an issue as discrimination under Title YII cases where if unlawful discrimination is found to be present, it is usually demonstrated to be present in a pattern or practice against a protected group as a whole. It found that the evidence established a prima facie case that a *610pattern and practice of age discrimination begins to appear at age 52 in its influence on layoff decisions and that the inference that age was a factor becomes stronger after age 55, increasing up to age 58, at which point it remains a steady influence until age 64. The trial court further found that the defendant’s statistical analyses were insufficient to defeat the prima facie case “that individuals ages 52 to 64 were adversely affected by layoff decisions in a pattern significantly different from other employees.”
The court noted that the issues were:
First, whether the defendant discriminated against employees from 40 to 65 years of age in selection of employees for termination in connection with the March 1973 layoff or reduction in force.
Second, whether defendant discriminated against Technical Staff members and Position Evaluation Plan employees from 40 to 65 years of age in defendant’s salary increase procedures (“stretch-out”).
Third, whether defendant discriminated against MTS employees from 40 to 65 in the salary allocation and distribution policies and practices which used age bands or groups during periods when these policies and practices were in effect.
Finally, whether during the periods relevant to these actions, defendant had discriminated against individuals from 40 to 64 years of age in recruiting and hiring policies.
•The third issue stated was withdrawn at the termination of the plaintiffs’ case by the Secretary.
The court explained that Sandia grew out of an engineering support division of the Los Alamos Scientific Laboratory in 1946. At that time a private corporation was formed, and Sandia was a prime contractor serving the Atomic Energy Commission, which became Energy Research and Development Administration. Western Electric is the parent corporation and does not receive a fee. All costs are, however, reimbursable and are funded by the federal government.
Sandia’s structure was also explained as consisting of a president, a number of vice presidents, directors, department managers and division supervisors. Those scientists and engineers engaged in the research and development activities are classified as MTS. SAT employees are involved in either technical or laboratory staff functions and set up in various classifications that are not particularly relevant to the present decision.
The population of Sandia Corporation totaled 8,200 in 1967 and declined to about 6,500 as a result of the 1973 layoff. In 1975, the company began research in non-weapon areas, chiefly energy programs. The court found that in the late 1960’s and early 1970’s, Sandia observed that the average age and length of service of the lab’s employees was increasing. It became concerned with the “profiles” of the MTS group. Believing that there was a threat to its technical vitality, it sought to employ young university graduates, particularly those with advanced degrees. The age guidelines given to recruiters were age 23 for a Bachelor’s Degree, 25-26 for a Master’s Degree, and 30 for a Ph.D. Those applicants who exceeded the age guidelines were not automatically eliminated, however. The recruiters were instructed to screen them closely so as to discover why they received their degrees beyond the guideline age. Recruiters would sometimes find individuals whose high grade point average was misleading due to a light academic course load. There were other explanations, for example, military service during the Vietnam conflict. Important to note is that there had to be an explanation as to why the applicant was delayed in receiving a degree. The recruiters insisted that age was not a criterion, but there were comments that age should not be considered as a detriment or that “we should overlook his age.” This same policy, the court found, was applied to the bulk of Sandia’s MTS hiring effort among unsolicited applications.
The court found that there was a group of exhibits designated “Bright Young Men *611Memos.” Actual hiring was engineered by the line organization interested in the applicant, not the recruiter or the personnel office. This was regardless of how the application was initiated. When a line organization requested that an individual be hired, a supervisor would write a laudatory memorandum and send it through channels. The court found that there were approximately 25 such memos received in evidence to cover the period 1965-1975. In the text of most of these exhibits there is a reference to the applicant as a bright or eager young man, and in some instances the writers observed that the individual is over 30 and then state why they are not concerned about his advanced age. The explanation is that these are words of description and not qualification. There are no comments as to whether these individuals were hired or rejected.
Plaintiffs’ expert in statistics, the court said, presented two hiring studies, neither of which added anything. The base data alone showed that few new hires came from the protected age group; that Dr. Spalding computed this to be 6%, but then he went on to compare this to data for engineers in nation-wide employment. The weight of this is affected by the fact that the 30% national figure included categories of engineers not used by Sandia, and it did not show availability for employment.
The court went on to describe the second hiring study of Dr. Spalding. The court found that they did not often recruit in the protected age group because they said older applicants were not available in significant numbers. They preferred to have recent formal training. At the same time, they did not close the door in an older applicant’s face. The inclination was to hire them only in instances where there was a vacancy. There was, however, little proof that established, so the court ruled, that there had been significant discrimination in this area of recruitment.
Finally, the court concluded that the Title VII cases with respect to hiring were not too helpful in an age discrimination case as to promotions and educational opportunity. The court was unable to find that these policies made any significant contribution to the end result.
As to the “stretch-out” system of salary increases, the trial court made a lengthy analysis and concluded that after weighing all of the evidence in the area, the plaintiffs had shown by a preponderance of the evidence that the “stretch-out” policy discriminated against plaintiffs in the protected age group on a class-wide basis. This was a method by which supervisors determined percentage increase of an employee’s salary and in doing so looked for the interval since his last raise. If the percentage and interval requirements were satisfied, the raise could be approved and effective immediately. If not, it was postponed to a time when the requirements were met. The judge found that this program was used in a very literal way and that, as a result, the “stretch-out” discriminated against workers in protected age groups. The court said that the burden of going forward with the evidence had shifted to Sandia to either rebut the prima facie case or present some other non-discriminatory reason for its salary practices.
What this amounted to was a rather strict use of criteria in giving an individual a raise, and it was found to work against persons within the protected age groups because of the more or less arbitrary postponement of wage increases.
The interlocutory decree was much too conclusory, extensive and detailed to suit Sandia. Sandia objected to the theory of analysis in the Interlocutory Opinion calling for examination of each year of age of each unit to ascertain whether an adverse effect had resulted in the layoff selections. The analysis was carried out for the six major employment categories at Sandia and for groups of those employee categories. It said that the approach lacked reasonable guidelines or standards. Strong objection was made to the fact that the trial court had concluded that a prima facie showing of discrimination had been made as to technical staff members for the six major categories. Sandia objected to the absence of reasonable guidelines or standards for the *612analysis, and a determination that an adverse impact had been sustained by one age group or one employee category was used as the basis for inferring discrimination in a different age group or dissimilar employee category. Another basis for the prima facie case was the receipt into evidence of what Sandia claimed was an assortment of exhibits detailing considerations for promotion and memorandum dealing with termination. It was said that the memoranda were too few, too insubstantial and too lacking in foundation to support the conclusion of age discrimination. It objected also to the fact that Sandia had not been allowed comparable discovery of the plaintiffs’ files.
Basically, the argument is that the burden of proof placed upon Sandia as a result of the determination that a prima facie case had been made out by the Secretary of Labor imposed an illegal or invalid burden.
SALARY ADMINISTRATION-THE TRIAL COURT’S FINDINGS
The findings made in these areas were for evidentiary value only, no relief was being sought by the Secretary. In fact, there was no allegation that the policies were in violation of the ADEA. This evidence indicated that written recommendations for promotion were often accompanied by considerations of age. The statistical evidence merely showed that in the MTS category, the mean age of those employees on roll increased over an eleven-year period (36.1 in 1964 to 41.4 in 1974), while the mean age of those promoted decreased over a ten-year period (36.3 in 1965 to 34.9 in 1974). (A low point in the mean age of those promoted was hit in 1971 at age 32.1.)
The trial court found the evidence as to Sandia’s promotion policy to be sufficient to cause “a reasonable suspicion” that age was a factor in promoting members of the technical staff to the first supervisory level; it declined, however, to comment on whether a prima facie case of age discrimination had been established from this.
The discussion of Sandia’s educational programs was brief. It included a year of college education toward a master’s degree, a doctoral studies program, a tuition support program for college courses taken outside of working hours and non-credit courses taught at Sandia’s laboratories. There was comment in the opinion that the full time college level programs were utilized by younger employees, but the court declined to comment on the evidence in the light of whether it did or did not establish a prima facie case in this area.
THE PERFORMANCE EVALUATION SYSTEM
Sandia employed a complex system for evaluating the performances of individual employees. This was largely for the purpose of administering salaries. It was shown to be used for such other purposes as selection for termination in the force reduction.
Beginning salaries of members of the technical staff were often determined by the use of several industry-wide compensation studies in order to determine amounts which competitive industries were paying to similar employees. These studies reflected the relationship between salary levels and either age or number of years since the employee received his final college degree. Based on these surveys, employees were assigned median beginning salaries and median full-proficiency salaries, with ranges allowed based on performance. Professional-level employees other than the technical staff, or PEP employees, were assigned scores based on practical ability, knowledge and accountability, with a 20% salary range above or below the dollar value attributed to the total number of points scored by the employee. The salary range was based on the employee’s performance rating within his level, such as Good-2 or Outstanding-6.
The system for increasing the salaries as the employee’s proficiency increased was performance evaluation for all non-graded (MTS and PEP) employees. Under this system, MTS employees were assigned a rank relative to other employees in their peer groups. This meant, prior to 1973, employees within their same age band: under 30, 31-36, 37-46, 47 and over. After 1973, San*613dia considered peer groups to be constituted from employees within the same range of years of experience in research and development: 0-5 years, 6-12 years, 13-20 years, and 20 and over years. Ranks were first assigned by an employee’s supervisor at the division non-supervised level, with each MTS employee in the division ranked on the basis of the supervisor’s judgment of the employee’s “job worth” and the employee’s performance in that job. The lists were then taken to the department level, where supervisors met with the department manager and the employees within the department were assigned ranks relative to their “peers” within the department. Also ranked at this level were the division supervisors, who were ranked with the non-supervisors (by the department managers). The process was then repeated at the vice presidential level. This resulted in a list of all MTS employees. This list was divided into eight equal numbers of employees, or octiles. Built into the process is a feature known as “zero sum balancing,” which requires one or more employees to move down in rank an equal number of steps as another employee moves up. The trial court found that non-supervisory MTS personnel in the protected age group tended to receive lower performance ratings because most supervisors had become supervisors as a result of being in the upper age bands and upper performance octiles. Employees were also grouped into salary octiles, and the selection of any particular MTS employee for a raise in salary was based upon the difference between his salary octile and his performance octile, the overall MTS salary structure and, prior to 1972, the employee’s “normal growth” for an MTS for his age and performance range. After 1972, the age factor was replaced by an experience factor.
PEP employees were ranked according to the same procedures, but were divided according to whether or not their jobs were supervisory. Also, comparisons were made according to jobs rather than age or experience. PEP employees were each required by a “forced distribution system” to be placed into a rating category: Outstanding, Good, Satisfactory, or Unsatisfactory. Within each category, except Unsatisfactory, were five levels, each representing a two percent change in salary level, so that salaries might range within 20% of an average, or control rate for a particular job depending upon an employee’s performance category and level within that category.
Sandia’s general salary structure is reviewed annually by a salary committee which can recommend adjustments. Each vice president is granted an allocation for salaries within his departments, to be used according to a division process which, prior to the force reduction, used age as a consideration, at least in the MTS job category.
Other salary policies which were used by Sandia, in addition to the performance evaluation system, included the rejection of salary increases of less than five percent as anti-motivational and possibly insulting, and the requirement that salary increases be granted no more frequently than annually, the system known as “stretch-out.”
Sandia argues vigorously that the company’s salary structure was a good faith effort to set forth criteria for establishing a salary scale within the business requirement of working with an appropriated budget which was in the process of being cut back. While admitting that the system was a subjective one, Sandia urges that performance evaluations of technical and scientific personnel cannot be done objectively, and points to the fact that the same system was approved by the same trial judge, who, in a 1976 Title VII action, had found that under Sandia’s circumstances, “subjective evaluation is virtually unavoidable.” Lawson v. Sandia Corp., No. 75-749-M (D.C.N.M. Dec. 8, 1976).
Further, Sandia urges, the trial court ignored the statistical interpretation that there was an absence of age discrimination, and the testimony of supervisors that individual employees were terminated for business reasons, thus embarking on the “dangerous course” described in Furnco Construction Corp. v. Waters, 438 U.S. 567, 98 S.Ct. 2943, 57 L.Ed.2d 957 (1978) of requir*614ing adoption of procedures considered to be “best” by the courts rather than those considered to be within the business judgment of an employer.
The trial court made several findings with regard to the performance rating system’s built in bias against the protected age group. One finding was that the system “is extremely subjective and has never been validated.” Comparing the system to that criticized in Brito v. Zia Company, 478 F.2d 1200, 1206 (10th Cir. 1973), “the evaluations were based on best judgment and opinion of the evaluators, but were not based on any definite identifiable criteria based on quality or quantity of work or specific performance that were supported by some kind of record.” Attempts by Sandia to reconstruct the ratings by use of objective criteria and asking supervisors to reclassify the affected employees were not accorded credence by the trial court, as the supervisors who re-rated the employees were aware of the approximate ratings given in 1973 under the subjective rating standard. The trial court noted that “[mjanagement’s concern about the increasing age of its staff, reduced hiring, new technical developments, and emphasis on recruiting and advancing young Ph.D.’s might not violate ADEA in themselves, but these policies and attitudes could easily be reflected in subjective performance ratings.”
The trial court then went on to illustrate the operation of these attitudes and policies within the performance rating system:
For example, if a supervisor wanted to reward a young Ph.D. or a promising young technician, he could raise the individual’s performance and salary position only at the expense of others in the unit because of the balancing requirement, [“zero sum balancing”]. Conversely, supervisors understood that raising the performance rating of an individual in the protected age group one or two steps might not be sufficient to get the person the minimum required increase because of the “stretch out policy.”
Commenting further on the interrelationships of other policies in use at the time of the reduction in force and the performance ranking system, the court stated:
There are also relationships between job assignments, motivation and performance ratings. Sandia admits that “stretch out” was discontinued because it was anti-motivational. Highly motivated employees are better performers. Sandia’s experts agree that challenging job assignments at other companies go to younger Ph.D.’s. In this case there were employees who were laid off because they had been in one line of research too long and were considered less versatile than younger employees. The evidence presented is not sufficient to prove or disprove the contention that at Sandia performance declines with age, but there is sufficient circumstantial evidence to indicate that age bias and age based policies appear throughout the performance rating process to the detriment of the protected age group.
Commenting on the documentary evidence, the trial court stated:
The Secretary introduced numerous exhibits showing that Sandia’s management and lower line supervision tended toward stereotyping older non-supervisory employees as unproductive and becoming technically obsolete. Management also believed that the future of the Lab depended upon “new blood,” that is, young Ph.D.’s. There is a factual explanation for this belief in the rapid changes in technology taking place at that time. The end result was an arbitrary generalization that older employees did not have the ability to keep pace with new developments. This “attitude” evidence corroborates the statistical evidence and supports an inference that age was a factor in selection for layoff.
It suggested that while the revised performance rating system might be a valid system for ranking Sandia’s MTS scientists, there was no business reason shown that non-scientists should also be affected by such a discriminatory policy.
Considering both the descriptive statistical and the documentary evidence and testi*615mony of the witnesses, the trial court concluded that Sandia’s evidence did not show that its actions were non-discriminatory toward the protected age group, that its policies were necessary to the safe and efficient operation of its business, or that the performance rating system was necessary throughout all of its employee classifications. The trial court did not appear to give much weight to the statistical inferences of either Drs. Prairie or Spalding. We agree with the trial court’s determination that Sandia failed to rebut the Secretary’s prima facie case. We agree also with the trial court’s determination that the plaintiffs had established by a preponderance of the evidence that Sandia had engaged in discriminatory conduct by selecting individuals in the 52 to 64 age range for layoff in the 1973 force reduction.
THE INDIVIDUAL CLAIMS
The several claimants whose individual problems were sponsored by the Secretary of Labor and are now prosecuted by the EEOC are being referred to as individual plaintiffs in the briefs. These were employees of Sandia who fell within the age limitation prior to the 1973 reduction in force. Sandia’s position as to each of these individuals is that the evidence is inadequate to justify the trial court’s awards. Specifically, attack is made on the injunction prohibiting Sandia from withholding payment of lost earnings and in some instances lost employer contributions to the company savings plan. Also claimed by Sandia is that each of the individuals failed to mitigate his damages. The trial court considered the case of each individual and granted relief in some instances. These will be considered hereinafter.
1. Loran G. Anderson
At the time of termination, Anderson was 56 years of age. His occupation was that of electrical engineer and he was classified as an MTS, a member of the technical staff. Sandia’s claim is that the prima facie case as applied to MTS employees was doubtful at best since only those in age group 56 to 63 were found to be affected by the company’s bias. It contends that the only evidence adduced in the remedial stage of the trial was a showing that the claimant was the second oldest employee in his division. No evidence, it is said, was adduced which challenged articulated responses bearing on Anderson’s termination. It is to be noted, however, that the department manager in Anderson’s department testified that Anderson’s employment was satisfactory. He said he relied upon the company’s performance rating system in making the selection. The trial court found that the referred-to performance rating system contained a built in age bias. There was a qualification in the court’s judgment as to the revised performance rating system as applied to MTS employees, but Anderson was judged by the old age-based system.
The remaining issue raised as to Anderson is that he did not seek to mitigate his damages. However, evidence was presented that he sought employment after notification of termination. He was awarded $29,044 back pay and $2,028 in lost savings. Also, Sandia was ordered to offer him reinstatement.
2. Vernon E. Baker
This employee of the photometries division, classified Staff Assistant Technical, was age 60. Sandia claims that in testimony offered, Baker’s manager told him that he should retire because of being 60 and because of having a heart condition, and that this was not sufficient in law to refute the articulated reasons that Baker had insufficient educational background to adjust to the current state of the art, high speed photography, and also that his job was discontinued.
The testimony by management was that Baker’s selection was the result of the tainted performance rating system. He held other jobs, but subsequently had two heart attacks and was unable to seek other employment. He was held to be entitled to $33,030 in lost earnings and $1,605 in savings contributions.
*6163. Frank M. Batchelor
This claimant was 57 years of age and his job was that of constructing batteries and also training for a calibration job in the Evaluation and Calibration Division. His classification was that of Staff Assistant Technical. The reasons given for terminating him were dissatisfaction with his job performance, his lack of technical education and unwillingness to be trained. He was said to be doing a “graded level job,” notwithstanding that he had a staff assistant classification. Following his termination, his job was abolished. His duties were assumed by a graded level employee. On the ground that no specific objection was made to these articulated reasons, Sandia urges that he was not entitled to relief. However, the performance rating system was used in his selection for termination. Batchelor did not present evidence in support of a claim for wages or savings. He was held to be entitled to be offered reinstatement.
4. Joseph S. Browning
This employee was 59 years old and was employed as a writer in the Weapons Evaluation Division. He composed engineering manuals. His classification was Staff Assistant, Laboratory. Sandia’s argument is that he presented no evidence as to his theory of his selection for discharge, except employee gossip, and that he did not challenge Sandia’s reasons; that he was the least effective of five writers in a staff which had to be reduced for economic reasons. The argument is that Browning had less formal training than the other four writers, and that the statistical evidence in this category was too small to be meaningful, that in the SAL category there were six employees of Browning’s age, two of whom were terminated. Of the 58-year-old employees in the category (one year younger than Browning), only one out of ten was terminated, and of 60-year-olds (one year older than Browning), three out of seven were laid off. At age 61, and above, only one out of 12 employees was terminated. These statistics, it is claimed, are insufficient to support the trial court’s finding of a pattern and practice within the job category.
The statistics cited by Sandia are, however, selective. Of the six employees Browning’s age, the evidence indicates that all six were selected for termination. Three were assigned to code category B (volunteered but selected), two (Browning included) in code category C (involuntary), and one in code category E (“saved”). (See Plaintiff Usery Exhibit 73 A.) Of seven 60-year-olds, two employees were selected for termination in addition to the three who were involuntarily terminated (none were “saved”), and out of the 12 employees over age 60, eight were selected for termination (BCD), none were “saved” and two volunteered, leaving only two such employees on roll. Browning’s department manager cited only the performance rating system for Browning’s selection for termination.
Following termination, Browning obtained employment with Sandia for a time. He suffered a heart attack and remained on Sandia’s rolls in an inactive status until he was age 65. The trial court awarded him $41,895 in lost earnings and $1,872 in savings.
5. John J. Colwell, Jr.
Mr. Colwell was age 56 at the time of the force reduction. He was a designer of parts for weapons carriers, was assigned to the Carrier Vehicles Division, and was, by job category, a Staff Assistant Technical (SAT). The articulated reason for his selection for termination was his limited technical background. Sandia urges that he made no rebuttal effort, even speculative, beyond the established prima facie case of a discriminatory pattern and practice of age discrimination. The supervisor who nominated Colwell for termination was deceased at the time of the trial, but there was evidence that the company’s performance rating system was a factor in his selection. Sandia urges that Colwell’s case is an example of court interference with the principle that an employer’s procedures in terminating employees need only be shown to be *617related to the achievement of a legitimate business purpose. See Furnco, supra. In this case, the purpose given was the continuation of the Upper Atmospheric Projects Department with a reduced number of employees. In order to reduce the number of employees, Sandia selected the employee with the least training. Sandia further urges that the statistics as applied to the SAT category of employees can create a different inference depending upon how the class is divided. It cites the fact that at Colwell’s age, 56, only three SAT’s out of 28, or 10%, were terminated, at age 55 two out of 21 (9%) were terminated, and at age 57, seven of 26 (27%) were terminated.
The statistics cited by Sandia take on a different aspect, however, when the evidence of employees selected for termination is considered rather than the number of employees terminated. At Colwell’s age, the combination of code B, C, D, & E employees is six out of 28 (21%). At age 57, it is nine out of 26 (35%). (Plaintiff Usery Exhibit 73 A). Moreover, the OGIVE chart for the SAT category for termination codes C and D, which was in evidence but not considered in the trial court’s approach to the statistical evidence, shows that the maximum difference between the probability that an employee was considered for termination by chance alone and the probability that he was considered for termination on the basis of age (all other factors being equal), occurs at about age 52. With the inclusion of code B employees, the maximum difference occurs at about the same age, and with the inclusion of the code E employees, the occurrence drops to age 45. (See Plaintiff Usery Exhibit 74.) It is not, of course, this court’s position to fix a statistical percentage in order for a finding to be sufficient to support an award. We must merely examine whether there was substantial evidence to support the finding. Furnco does not require abdication of the substantial evidence test in employment discrimination cases.
Colwell sought and obtained employment following termination. His entitlement was fixed at $30,842 lost earnings. No savings contributions were found to be owed. He was also held to be entitled to an offer of reinstatement.
6. Carl E. Drew
This employee was age 58 and a Technical Staff Assistant (TSA) when terminated from the Process Metallurgy Division of the Metallurgy Department. He served on the component welding staff. He said that he had no direct knowledge of the reasons for his termination other than the general knowledge that employees over age 50 with over 20 years’ tenure were likely to be selected in the force reduction program. His department manager testified that the decision had been made to operate the Metallurgy Division with two “professionals” instead of its current three. The company followed a different procedure in the division than the company-wide procedures used elsewhere in selecting employees for layoff. Drew’s case was considered in relation to other “professionals,” including a doctoral level staff member, even though there was evidence that the division director had instructed that the selection be based upon the organization’s ability to survive with the least damage. Drew contended that under this standard, he should have been tested against non-professional employees in the division. Drew’s department manager testified that he relied partially on the company’s performance evaluation system in making the selection of Drew for termination.
Drew held several jobs after termination. He sought and obtained more substantial employment in another laboratory, which he left at age 62. He was awarded $37,268 lost earnings, plus reinstatement.
7. Mark Elich
Mr. Elich was an instructor in the Weapons Training Division, classified as a Member of Laboratory Staff (MLS). He was age 53 when selected for termination. He had a Bachelor of Science degree in Electrical Engineering and a Master’s degree in Education, which was admittedly the proper technical background for his job. The pur*618ported reasons for his discharge were: reluctance to travel (which he denied), and a suggestion of a lack of initiative. He was selected for termination based on the company’s performance evaluation system and his alleged low performance rating. Sandia argues that the sampling was too small in this category. This is not supported by the OGIVE statistics.
Elich applied for several job openings in his field, but only found part-time work as a substitute schoolteacher. He was awarded lost earnings in the amount of $43,835, savings contributions in the amount of $1,317 and an offer of reinstatement.
8. Roy Ewing, and 9. Leon Filvin
The trial court considered the cases of both of these individuals together.
Both were Staff Assistant Laboratory (SAL) employees, working as design draftsmen. Although they worked in different departments, they were responsible to the same superior and were both selected for termination according to the same standards. Ewing was age 57 and Filvin was 60 at the time of the layoff. Sandia attacks the size of the category as being too small to reflect a true picture. The reasons given for their terminations were related to a 1968 reduction in the number of employees employed in drafting, which, because of rapid technological changes, created more difficult work for fewer draftsmen. Both men were placed at the bottom of performance ranking lists which had been prepared over a three-year period. Their superior testified that these lists were used exclusively in the selection process. Both employees were considered deficient in technological training, with Filvin having had no formal technical training since the 1930’s and having an alleged “disdain if not hostility” to advances in the state of the drafting art, from copiers to computers. Ewing had not taken formal training for ten years except for a correspondence course. Filvin challenged whether knowledge of the technological changes was necessary to the performance of his duties. Ewing challenged his ranking as compared to younger, newer employees, whom, he claimed, came to him for instruction when needed. Both made considerable effort to show that the reasons articulated for their layoffs were pretext, and the trial court found that Sandia failed to overcome the presumption of age discrimination.
Ewing became employed after his termination, and no monetary sums were found to be due him. He was found, however, to be entitled to an offer of reinstatement.
Filvin sought employment for some time, but became disillusioned when he was rejected due to his feeling that he was too old to be hired. He was found to be entitled to lost earnings in the sum of $40,089 and lost savings contributions in the sum of $1,552.
10. Donald Fifield
Fifield was age 58 and was the oldest employee in the Mobile and Remote Ranges Division of Sandia. He was engaged in arranging and conducting rocket tests at various test sites. His job category was that of Laboratory Staff Assistant (LSA). His selection for discharge was disputed by his immediate supervisor, and was said to be based on his lack of technical background, since Fifield had only a high school education. Fifield alleged that a technical background was unnecessary to his job. This allegation was admitted in Sandia’s brief. Sandia’s claim was that the Nuclear Test Ban Treaty signed by the United States resulted in a reduction in nuclear tests from three per month to three per year. Sandia’s further claim was that statistical evidence upon which the prima facie case for LSA employees was based was too small to support any inference, and, in fact, moved “to strike any presumption or prima facie case” prior to the taking of testimony in Fifield’s portion of the case. Sandia urges that the trial court took this motion under advisement but failed to rule on it, except in its finding that such presumption was not overcome. The statistics relating to the LSA category show a total of eight employees of Fifield’s age. Two of these (25%) were involuntarily terminated. At age 57 (one year younger than Fifield), *619there were five employees, none of whom were involuntarily terminated (0%), and at age 59 (one year older), there were five employees, two of whom (40%) were involuntarily terminated. By comparison, with the addition of those employees in the LSA category who were selected for termination, the figures become 20% (one in five) at age 57, 38% (three in eight) at age 58, and 60% (three in five) at age 59. No argument going to the weight of the size of the statistical sample inherent in the number of LSA employees or any portion thereof was made.
Fifield sought other employment after termination and became self-employed for a period without much success. He was engaged in the administration of a relative’s estate for a time, which factors were considered in the trial court’s finding that he was entitled to lost wages in the sum of $33,598 and employer’s savings contributions in the sum of $617.
11. Mark B. Gens
Mr. Gens was the eleventh employee to be considered in the Secretary’s case in the October 1978 trial.7 He was 58 years of age when selected for termination. He was allowed to remain on roll until he became 60 in 1975. He had been previously employed by Sandia as a packaging engineer in the Criteria and Heat Transfer Division of Sandia, classified as Member of the Technical Staff (MTS). His section was dissolved and he was transferred to Sandia’s Environmental Data Bank. This section considered the environmental stresses to which Sandia’s products would be submitted.
Gens pointed to the fact that he felt his age was a factor in his selection for termination because no older employees in his category were “snatched back” in the reconsideration. He also pointed to the fact that one reason advanced for his termination, the decision to reduce the staffing of the data bank, was one which was later rescinded when he was allowed to remain on roll. Sandia admits that the data bank was later increased in manpower, but claims that this was a result of Sandia’s entry into the field of energy development, a decision made at a later point in time. Sandia claims that Gens was the logical person to terminate in a hotly disputed decision to reduce the size of the data bank staff. Sandia does not, however, explain why Gens was the logical employee to lay off, but relies on the evidence which, Sandia contends, established overwhelmingly that Gens was not selected by a process motivated by age consideration. Gens testified that he believes something was said about his “lack of flexibility” in connection with termination.
*620Gens was selected for termination by his department manager who testified positively about all of Gens’ qualities for the job. He further said that Gens’ job was determined to be the least important function in the department. In addition to Gens, some young employees were also terminated. The division supervisor, directly under the department manager, disagreed with the decision to terminate Gens, and an unsuccessful effort was made by people at the division level to rescind the decision. He also confirmed that Gens was told he lacked “flexibility.”
Gens also testified that much of his feeling that he had been discriminated against because of his age arose from the fact that older employees who had been terminated were not included in the rescission of that action, while younger employees were. Gens’ department manager, who actually made Gens’ selection for termination, testified that the company’s performance ranking system was not a consideration in the decision. That system showed Gens in the lowest performance octile. It also showed that all four of the employees in Gens’ department who were ranked in the eighth, or lowest, octile were over 40 years of age, and all employees in the second (next to highest) octile were 35 years of age or younger. However, in the actual termination, some older low-ranked employees in the department were retained at the expense of younger employees at a higher rank.
Sandia vigorously urges that Gens’ case is illustrative of the difficulty of reducing Sandia’s staff on the basis of business considerations, and that Gens was merely the most expendable employee in the most expendable job. The fact that there was intracompany disagreement over these issues and the fact that the job was later given a higher priority, it is urged, do not detract from the legitimacy of the decision to terminate Gens, but, rather, show the high degree of consideration and good faith which went into the selection and the subsequent efforts to obtain reconsideration.
True, testimony involving Gens’ selection (for termination) shows that much consideration was given to the decision, and neither Gens’ age nor age-tainted performance ranking entered into the consideration. There is substantial evidence that Gens bore a low, although probably undeserved, performance rating and that the reason given to him for his termination included his lack of “flexibility,” which was explained in terms of his “educational background and his previous job experience.” Gens was awarded the sum of $44,013 as lost earnings, $674.00 as lost savings contributions and was ordered to be offered reinstatement.
The trial court was free to disbelieve the testimony that the age-biased performance ranking system was ignored in Gens’ case, particularly in light of that system’s widespread use in the company’s policies concerning promotion, salary increases, and in other departments, selection for termination. There is, therefore, substantial evidence to support the trial court’s conclusion. Sandia did not prove that Gens was not an identifiable victim of the company’s discriminatory procedures in terminating employees in the reduction in force. See Teamsters v. United States, 431 U.S. 324, 97 S.Ct. 1843, 52 L.Ed.2d 396 (1977).
The company was able to make a relatively strong showing in the Gens case. It demonstrated an extensive effort to find a place for him in some other endeavor and claims that this showed its good faith. We recognize this, but it does not gainsay that the underlying cause continued to be his age together with the continuing need to reduce the number of employees. The trial court’s action is supported by the statistics and the circumstances.
A. The issue whether the trial court erred in determining that the Secretary of Labor had established a prima facie case of age discrimination.
Sandia contends that the district court’s determination that a prima facie showing of age discrimination had been made in general with respect to the employees between the ages of 52 and 64 for layoff was *621erroneous because it was based, so it is argued, upon a statistical analysis, the underlying theory of which was to examine each year of age as a separate unit in ascertaining whether an adverse impact resulted from the layoff selections. It is said that the analysis by the district court was unwarranted by the evidence; that it lacked reasonable guidelines or standards; and that an adverse impact sustained by one category was used as the foundation for inferring discrimination in a different age group or dissimilar employee category. It is also argued that the determination that a prima facie showing had been made had placed an unreasonable burden upon Sandia.
The government responds that the court correctly applied the burdens of proof as enunciated by the Supreme Court in McDonnell—Douglas Corp. v. Green, 411 U.S. 792, 93 S.Ct. 1817, 36 L.Ed.2d 668 (1973); International Brotherhood of Teamsters v. United States, 431 U.S. 324, 97 S.Ct. 1843, 52 L.Ed.2d 396 (1977); and also Furnco Construction Corp. v. Waters, 438 U.S. 567, 98 S.Ct. 2943, 57 L.Ed.2d 957 (1978) and Board of Trustees of Keene State College v. Sweeney, 439 U.S. 24, 99 S.Ct. 295, 58 L.Ed.2d 216 (1978). The government argues that Teamsters rejected the Sandia view that the trial court, following the testimony offered by the defendant in a discrimination case, has to accept the testimony on its face without judging its credibility or considering it in relationship to the plaintiffs’ case. The government maintains that the trial court followed the three-step procedure that was prescribed in Teamsters, wherein it was said that, first, the plaintiff proves a prima facie case, if possible; second, defendant offers rebuttal; and, third, the plaintiffs offer evidence of pretext. All of this occurs within the initial or trial stage of a pattern or practice suit, and at that point the government “establishes a prima facie case that such a policy existed.” 431 U.S. at 360, 97 S.Ct. at 1867. The burden then shifts to the employer to defeat the prima facie case. If the employer fails to rebut the inference that arises from the prima facie case, the trial court may then conclude that a violation has occurred and determine the appropriate remedy. “Without any further evidence from the Government, a court’s finding of a pattern or practice justifies an award of prospective relief.” 431 U.S. at 361, 97 S.Ct. at 1867.
The contention of Sandia that the trial court failed to make a finding that Sandia had engaged in discriminatory practices is not correct. In its Interlocutory Opinion given October 20, 1977, the trial court said:
I hold that the plaintiffs have established by a preponderance of the evidence that individuals in the 52 to 64 age range were selected for layoff in a pattern which proves that Sandia has engaged in discriminatory conduct.
It cannot be said that this was not a determination that a prima facie case was established.
As noted above, the same contention was advanced in International Brotherhood of Teamsters v. United States, supra. There the Supreme Court reviewed the procedure for order of proof in a pattern and practice lawsuit under Title VII. The Court in Teamsters made clear that the McDonnell-Douglas Corp. doctrine of procedure is not an inflexible formula applicable to discrimination cases. 431 U.S. at 358, 97 S.Ct. at 1866. The Court reviewed its prior decision in Franks v. Bowman Transportation Co., 424 U.S. 747, 96 S.Ct. 1251, 47 L.Ed.2d 444 (1976), which was a Title VII class action in which the holding was that the plaintiffs had established a prima facie case of discrimination against individual class members by demonstrating the existence of a discriminatory hiring pattern and practice. 424 U.S. at 772, 96 S.Ct. at 1267-68. Thereafter, the burden shifted to the employer to prove that employees who reapply were not in fact victims of previous hiring discrimination. 424 U.S. at 772, 96 S.Ct. at 1267-68. The Teamsters opinion also held that proof of a discriminatory pattern and practice creates a rebuttable presumption in favor of individual relief. The Teamsters opinion set forth at some length the stan*622dards to be employed in a pattern and practice suit:
The plaintiff in a pattern-or-practice action is the Government, and its initial burden is to demonstrate that unlawful discrimination has been a regular procedure or policy followed by an employer or group of employers . . . . At the initial, “liability” stage of a pattern-or-practice suit the Government is not required to offer evidence that each person for whom it will ultimately seek relief was a victim of the employer’s discriminatory policy. Its burden is to establish a prima facie case that such a policy existed. The burden then shifts to the employer to defeat the prima facie showing of a pattern or practice by demonstrating that the Government’s proof is either inaccurate or insignificant.
If an employer fails to rebut the inference that arises from the Government’s prima facie case, a trial court may then conclude that a violation has occurred and determine the appropriate remedy. Without any further evidence from the Government, a court’s finding of a pattern or practice justifies an award of prospective relief.
When the Government seeks individual relief for the victims of the discriminatory practice, a district court must usually conduct additional proceedings after the liability phase of the trial to determine the scope of individual relief. The petitioners’ contention in this case is that if the Government has not, in the course of proving a pattern or practice, already brought forth specific evidence that each individual was discriminatorily denied an employment opportunity, it must carry that burden at the second, “remedial” stage of the trial. That basic contention was rejected in the Franks case.
The proof of the pattern or practice supports an inference that any particular employment decision, during the period in which the discriminatory policy was in force, was made in pursuit of that policy. The Government need only show that an alleged individual discriminatee unsuccessfully applied for a job and therefore was a potential victim of the proved discrimination. As in Franks, the burden then rests on the employer to demonstrate that the individual was denied an employment opportunity for lawful reasons.
431 U.S. at 360-62, 97 S.Ct. at 1867.
Any nondiscriminatory justification offered by the company will be subject to further evidence by the Government that the purported reason for an applicant’s rejection was in fact a pretext for unlawful discrimination.
431 U.S. at 362 n. 50, 97 S.Ct. at 1868.
Sandia’s position is that the establishing of a discriminatory pattern and practice shifts the burden of going forward with the evidence, but does not shift the burden of persuasion from the plaintiff. Sandia maintains that the proof of a pattern or practice creates a mere inference and that the district court erred by giving too much effect to it. Sandia argues that the inference fades away automatically once the employer offers a non-discriminatory explanation.
Teamsters, however, holds that in pattern and practice cases the burden of proof in the sense of the burden of persuasion shifts to the employer once a discriminatory pattern and practice has been shown. It is not necessary here to enunciate a conclusion as to whether there is indeed a shifting of the burden of proof. In this case the distinction is of no importance because the burden of persuasion was maintained by the plaintiffs, and it must be remembered that this cause was tried to the court. It is not an appeal from a directed verdict. This same argument was made at the district court level, and the judge there said:
In . concluding [that Sandia violated the ADEA with respect to 15 of the 17 individual claimants] the burden of proving the case, the burden of persua*623sion as to the plaintiffs’ claims, rested at all times on the plaintiffs. In consideration of all of the evidence, however, whether that evidence was introduced at the first phase of the trial or introduced at the second or third phase of trial, I concluded that the defendant’s articulated reasons regarding its treatment of these first 17 individuals were in large part not believable, not supportable, otherwise not legitimate, and in fact, were pretext for unlawful discrimination.
I do not believe that any case on discrimination requires me to accept as defeating a plaintiff’s claim any or every proposition, statement, or ground put forth by the defendant. In the present suit I have concluded, after hearing defendant’s explanations and reasons, that many of the explanations and reasons are neither legitimate nor credible. In these instances I found for the plaintiff.
It is thus shown that the district court did treat the Sandia justifications and was of the opinion that “defendant’s explanations and reasons . . . are neither legitimate nor credible. In these instances I found for the plaintiff.” Thus, it is apparent that the district court discounted the reasons offered by Sandia as justifications for the layoffs simply because Sandia’s reasons were not credible. The court did not impose a disproportionate burden on Sandia during the second trial. It took notice of the fact, however, that in order to rebut the presumption or inference created by the initial finding, the employer must advance nondiscriminatory reasons for its actions. The court was of the opinion that Sandia failed to do this.
A further contention of Sandia is that at the second remedial stage of the proceedings, individual claimants must present rebuttal evidence of pretext to counter the employer’s justifications and if they do not, the court must find for the employer. However, this is not correct. This order of proof does not have to march in lockstep so long as the evidence is presented and the parties are given full opportunity to present their evidence. Cf. Whack v. Peabody & Wind Engineering Co., 595 F.2d 190, 193 n. 10 (3rd Cir. 1978).
We conclude that the district court did not err in its assignment of the burden of proof in this ease. The court’s finding of discriminatory pattern and practice is supported by the evidence, not only the statistics, but the testimony of witnesses to the practices of Sandia, and the result of the court’s findings was that Sandia was proven to be a violator following the first stages of the proceedings. See International Brotherhood of Teamsters v. United States, supra, 431 U.S. at 359 n. 45, 97 S.Ct. at 1867. In considering that the individual claimants were affected by Sandia’s discriminatory practices, and considering that the reasons given for termination were not credible, the individual claimants are entitled to relief.
B. The question whether the trial court’s finding that there was a violation of the ADEA was clearly erroneous.
We have covered much of the argument under this heading. Thus, we have considered the three-stage procedure for this kind of action and have concluded that the trial court did not err in the procedure that it followed.
Sandia maintains that once it gave some kind of explanation following the prima facie showing on the part of the Secretary, its burden was completed and somehow the showing made by the government in its case-in-chief evaporated. This, of course, is not so. The record as a whole is certainly the final test of sufficiency of the evidence.
As to the individual claimants and the sufficiency of the evidence as to them, surely the case-in-chief of the Secretary is entitled to full consideration. The statistics provide an extraordinarily strong circumstantial basis favoring not only the general showing on the part of the Secretary, but also the specific cases of the claimants, but that is not the only evidence in the record. It cannot be denied that the factors of the numerous exhibits together with the direct testimony of persons having actual knowledge of Sandia’s preoccupation with the *624“problem” of old age, have been pervasive in all of the evidence. Indeed, the decision to terminate a large percentage of the employees in 1973 gave rise to an occasion for getting rid of the old and bringing in the new.
We have also commented briefly in the opinion concerning the individual problems and on some of the evidentiary matters connected with each individual claimant. This, taken with the circumstantial and direct evidence mentioned, we regard as substantial and extraordinarily strong evidence in support of the Secretary’s case, including that on behalf of the individuals, that is, Anderson, Baker, et al.
This is a most unusual case in that the age factor was ever present. The governing group in Sandia was, it seems, always concerned with utilizing the retirement device in order to rid the company of workers who were somewhat advanced in age. It is unusual also because of the impersonalism and inconsiderateness toward human dignity-
C. Whether the district court erred in its award of damages for violation of ADEA and in ordering offers of reinstatement to the terminated employees.
Sandia contends that the trial judge erred in the following respects:
First, in its failure to offset unemployment compensation, layoff allowance and vacation pay which was received by certain claimants.
Second, that the court erred in concluding that the claimants had made reasonable efforts to mitigate damages.
Third, that the court erred in making offers of reinstatement to claimants under the age of 65 at the time of judgment.
Most of the foregoing contentions are alleged abuses of discretion.
We first consider whether there was error in failure to offset the earnings. Should unemployment benefits be offset against amounts received as back pay? The trial court said “No.” It relied on N. L. R. B. v. Gullett, 340 U.S. 361, 71 S.Ct. 337, 95 L.Ed. 337 (1951), in arriving at its conclusion that such benefits were not to be deducted. In Gullett, the Supreme Court held that in awarding back pay, in accordance with the National Labor Relations Act, the Board did not abuse its discretion. It said:
Such action may reasonably be considered to effectuate the policies of the Act. To decline to deduct state unemployment compensation benefits in computing back pay is not to make the employees more than whole, as contended by respondent. Since no consideration has been given to collateral losses in framing an order to reimburse employees for their lost earnings, manifestly no consideration need be given to collateral benefits which employees may have received.
340 U.S. at 364, 71 S.Ct. at 339.
The Fifth Circuit followed Gullett in affirming the trial court’s refusal to deduct unemployment benefits in an ADEA case, Marshall v. Goodyear Tire & Rubber Co., 554 F.2d 730 (5th Cir. 1977). It was held that the trial court had not abused its discretion in denying the request to deduct such benefits. It was noticed in that case that the trial judge is authorized to grant such relief as may be appropriate, including the enforcement of an employer’s liability for lost wages. Because of the breadth of the trial court’s authority under Section 7(b) of ADEA, 29 U.S.C. Section 626(b), it was concluded that the district court was correct in its determination.
A number of district courts have followed this reasoning and have refused to offset unemployment benefits on the basis that such benefits are collateral to back wages and are awarded by the state in furtherance of a separate social policy. Pedreyra v. Cornell Prescription Pharmacies, Inc., 465 F.Supp. 936 (D.Colo.1979).8 See also Abron *625v. Black & Decker Mfg. Co., 439 F.Supp. 1095 (D.Md.1977); Tidwell v. American Oil Co., 332 F.Supp. 424 (D.Utah 1971).
This court's decision in Local Lodge No. 790, I. A. M. A. W. v. Champion Carriers Inc., 470 F.2d 744, 745 (10th Cir., 1972), construed the language of a labor arbitrator’s award for wrongful discharge which granted full back pay less any amount earned from other employment. The court cited N. L. R. B. v. Gullett Gin Co., supra, and stated that “Clearly the language of the award does not allow deduction of unemployment compensation benefits.” Id. at 745. This Tenth Circuit decision is based upon a labor arbitrator’s order rather than a statute, but it does have some similarity in that it involves the deduction of unemployment benefits.
Sandia contends that the refusal to allow a setoff is error because it allows claimants to receive a windfall recovery by receiving back pay and unemployment compensation covering the same period. The answer to this is that the unemployment compensation is purely a collateral source and is peculiarly the property of the claimant. It would be unfair to give Sandia the benefit of it in these circumstances.
There are decisions which have held that no abuse of discretion occurred when the trial court offset unemployment compensation benefits from a back pay award. E. E. O. C. v. Enterprise Association Steamfitters Local No. 638, 542 F.2d 579 (2d Cir. 1976), cert. denied 430 U.S. 911, 97 S.Ct. 1186, 51 L.Ed.2d 588 (1977); Ostapowicz v. Johnson Bronze Co., 541 F.2d 394, 401 (3rd Cir. 1976), cert. denied 429 U.S. 1041, 97 S.Ct. 741, 50 L.Ed.2d 753 (1977) (affirming without discussion the district court’s calculation of back pay award which offset unemployment insurance in a Title VII case); Satty v. Nashville Gas Co., 522 F.2d 850 (6th Cir. 1975), vacated and remanded on other grounds, 434 U.S. 136, 98 S.Ct. 347, 54 L.Ed.2d 356 (1977) (general approval without discussion of district court’s deduction of unemployment benefits in a sex discrimination case under the Equal Employment Opportunity Act); Bowe v. Colgate Palmolive Co., 416 F.2d 711, 721 (7th Cir. 1969) (This was a Title VII case in which the court held that the deduction of unemployment compensation was a valid exercise of the trial judge’s discretion.)
The court in E. E. O. C. v. Enterprise Assoc. Steamfitters Local 638, supra, recognized the analogy to the N.L.R.B. policy, approved by the Supreme Court in Gullett in refusing to allow deduction of collateral benefits such as unemployment compensation from back pay awards, but affirmed the trial court’s order that public assistance be deducted from a back pay award in a Title VII case. Also recognized was that the weight of common law authority is that collateral amounts are not deducted from a damage award even where payment has been received from a fund supported in part by contributions by the defendant, citing 2 F. Harper & F. James, The Law of Torts Section 25.22 at 1343 n. 1 (1956). Nevertheless, the court, “as a matter of policy,” agreed “with the rulings of other circuits which have held it not an abuse of discretion to deduct sums received from collateral sources such as unemployment compensation.” It stated:
We see no compelling reason for providing the injured party with double recovery for his lost employment; no compelling reason of deterrence or retribution against the responsible party in this case; and we are not in the business of redistributing the wealth beyond the goal of making the victim of discrimination whole.
542 F.2d at 592.
The circuit court cases which give approval to the crediting of unemployment compensation benefits are largely based on upholding the discretionary decision of the district court. No one of them shows a marked preference for this doctrine. Even though this approach is somewhat questionable, it is in accordance with the Supreme *626Court’s decision in N. L. R. B. v. Gullett Gin Co., supra.9
The deduction or offsetting of unemployment benefits may well result in a windfall to the employer. He finds himself in a position where he is not responsible for the payment of the illegally withheld back pay and then offsetting it with unemployment benefits by the government, which is unjust enrichment except to the extent that employers make contributions to the fund.
Sandia does not make an argument based on the employer’s contribution. Sandia also argues that vacation allowance received by claimants at the time of discharge should have been charged against back pay awards, but the vacation allowance compensated claimants for vacation time which had accrued under Sandia’s policies, but which had not been used. Inasmuch as this vacation time had been earned, there is no good reason for its being an offset where back pay is awarded. See Pettyway v. American Cast Iron Pipe Co., 494 F.2d 211 (5th Cir. 1974).
A further question is whether it was error not to recognize layoff allowances received by various claimants as an offset. Sandia had a system which called for a termination payment measured by the length of time the employee had been working for Sandia. Some of the employees elected to defer their termination for a period of time in order to reach a date critical to enhanced pension benefits. In such cases the layoff allowance or termination pay was reduced or eliminated according to the period of time that the employee remained on the payroll. Sandia’s theory is that the layoff allowances are an incident of termination which would not have been received by the employee and thus they are duplicative of the back pay awards for wrongful termination.
Some courts have recognized the validity of Sandia’s argument. See Laugesen v. Anaconda, 510 F.2d 307 (6th Cir. 1975); Coats v. National Cash Register Co., 433 F.Supp. 655 (W.D.Va.1977); Combes v. Griffin Television, Inc., 421 F.Supp. 841 (W.D.Okla.1976).
The claimants maintain that it was a proper exercise of discretion by the trial court not to offset the allowance. They put it on the same basis as the principle governing the offsetting of unemployment compensation. They claim that it is a collateral benefit which compensates claimants for losses not covered by the back pay award and therefore is not a double recovery. The trial court gave no reason for denying the offset. However, in the related case, Mistretta, No. 79-1021, the court stated that “The reasonable interpretation of a layoff allowance is compensation for the costs of relocation and the expense of finding other employment.” Back pay awards seek to make whole discharged employees for their lost wages and not for extra expenses. The claimants assert that since it is the very fact of termination that earned them the right to a layoff allowance, the court’s refusal to subtract these amounts was not an abuse of discretion.
However, in this area the cases support the offsetting of severance pay since it is a payment made wholly by the employer; thus it is not a collateral benefit and moreover would not have been made if termination had not occurred and, therefore, goes beyond the damages necessary to place the claimants in a whole position.
The claimants say that the trial court had the discretion to not offset employment compensation benefits on the theory that they are collateral payments, and indeed they are. It is said that it might be appropriate to apply the same reasoning to the layoff allowance. This appears, however, to be open to question because it is not really a true collateral benefit. This would be especially apparent in the case in which reemployment is ordered.
*627Due to the fact that this is not an item that comes from a third person and is therefore plainly a collateral benefit, we are disposed to conclude that this is not governed by the same principle that applied to the unemployment payments. This was social insurance payable by the state and the beneficiary was the claimant.
The final question is whether there was a duty, and violation of such duty, to mitigate damages by the employees. The trial court reduced all back pay awards by amounts received from other employment. Back pay was not awarded claimants for those periods of time when they were found to be unavailable for other work or were not seeking other employment. Two claimants were found to have failed to prove any monetary damages from the termination and were awarded no back pay. As to the claimants who were actually awarded back pay, the court found that reasonable efforts had been made to mitigate. Sandia claims that these findings were not supported.
Unquestionably, wrongfully discharged claimants have an obligation to use reasonable efforts to mitigate their damages. The only issue is whether the court’s finding that reasonable diligence had been exercised was clearly erroneous. See United States v. Smurthwaite, 590 F.2d 889 (10th Cir. 1979).
In United States v. Lee Way Motor Freight, Inc., 625 F.2d 918, (10th Cir., 1979), we said: “A claimant is required to make only reasonable exertions to mitigate damages, and is not held to the highest standards of diligence. It does not compel him to be successful in mitigation. It requires only an honest good faith effort.” 625 F.2d at 938. The court said further that “once a violation has been demonstrated and back pay has been awarded, the employer has the burden of showing that the discriminatee did not exercise reasonable diligence in mitigating damages caused by the employer’s illegal action.” Id. at 937. See also Sias v. City Demonstration Agency, 588 F.2d 692, 696 (9th Cir. 1978); Kaplan v. International Alliance of Theatrical & Stage Employees, 525 F.2d 1354, 1363 (9th Cir. 1975); Sprogis v. United Airlines, Inc., 517 F.2d 387, 392 (7th Cir. 1975); Sparks v. Griffin, 460 F.2d 433, 443 (5th Cir. 1972); Hegler v. Board of Education, 447 F.2d 1078 (8th Cir. 1971).
In order to satisfy the burden, the court in Sias v. City Demonstration Agency, supra, at 696 held, “the defendant must establish (1) that the damage suffered by plaintiff could have been avoided, i. e. that there were suitable positions available which plaintiff could have discovered and for which he was qualified; and (2) that plaintiff failed to use reasonable care and diligence in seeking such a position.”
Sandia has not been able to bring forward any evidence indicating that other comparable positions were available for the claimants. Furthermore, the trial court’s finding that reasonable diligence was exercised is supported by the record and is not clearly erroneous. We conclude that Sandia has not satisfied its burden to show that reasonable efforts to mitigate damages were not made.
Sandia’s final point is that it was abuse of discretion for the court to order offers of reinstatement in the classification which had been occupied. This was limited to persons under 65 years of age at the time of trial. It was pointed out that there were three persons who had informed the Department of Labor that they would not accept reemployment with Sandia even if the same were offered. It is difficult, however, to cite prejudice in the court’s order to make offers of reinstatement to claimants who are not going to accept them. Such statements, in any event, do not constitute a waiver of reinstatement when such statements were made prior to the offer of the employment in an action pursuant to the National Labor Relations Act. Heinrich Motors, Inc. v. N. L. R. B., 403 F.2d 145 (2d Cir. 1968). There the court approved the N.L.R.B.’s statement that such a conversation by an employee may reflect only a momentary state of mind that is subject to change and “prior to an offer of reinstatement, such statements are in the nature of answers to a hypothetical question; and the *628discriminatee’s expression may have been made in the heat of dissatisfaction with his treatment by [the employer.]” Id. at 150. The court further said:
Moreover, there is no indication that the [employer] knew that [the plaintiff] did not desire reinstatement or that it relied upon the alleged statement in delaying its reinstatement offer. The reinstatement requirement has the dual purpose of protecting the discharged employee and demonstrating the [employer’s] good faith to its other employees. Local 883 U. A. W., etc. (Kohler Co.) v. N. L. R. B., 300 F.2d 699, 703 (1962).
Id. at 150.
The conclusion was that the statements attributed to the plaintiff could not be justifiably regarded as a basis for excusing the employer from its duty to offer reinstatement.
This reasoning can be applied readily to the case at bar.
# * #
In summary, age discrimination is shown to have been predominant in the entire case. For a long period of time the supervisory element in this company was sensitive to the advanced age problem. It was almost what amounted to a presumption that anyone who reached a certain age was less able to produce in accordance with their standards. There was no doubt that once the 1973 reduction in force policy came out that the first outlet for reduction was in the area of persons between the ages of 52 to 65.
That the supervisors were highly sensitive to this age question is shown by their policies with respect to hiring. Indeed, there was reluctance to hire persons with advanced degrees if they had obtained these degrees a few years subsequent to the obtaining of their baccalaureate degrees. In other words, some reason had to be provided for this delay. We need not recount the other circumstances which evidence this policy. It is sufficient to say that one will seldom find a case in which the statistics support the findings and conclusions and the direct evidence corroborates the statistics.
The work of the trial judge was impressive. Not only did he devote a great amount of time and attention to the details of this case, he became, in addition, well educated in the signs that indicated the underlying policies of the Sandia organization.
Therefore, we unhesitatingly affirm the judgments of the district court, with the exception that the actions are remanded to the district court for the purpose of allowing layoff allowances paid by the appellant as an offset against the damage awards.