Important legal notice

61997C0309

European Court reports 1999 Page I-02865

Opinion of the Advocate-General

I - Introduction

1 In this reference for a preliminary ruling under Article 177 of the EC Treaty, the Oberlandesgericht Wien (Austria) seeks replies to seven questions concerning the interpretation of Article 119 of the EC Treaty and Council Directive 75/117/EEC of 10 February 1975 on the approximation of the laws of the Member States relating to the application of the principle of equal pay for men and women. (1) The questions referred to the Court mainly concern the interpretation of the concepts of `equal work' and `same job', and the identification of the groups to be compared in order to ascertain whether there is indirect discrimination between men and women with regard to pay.

II - Legal context

A - Community provisions

2 The first paragraph of Article 119 of the Treaty provides:

`Each Member State shall during the first stage ensure and subsequently maintain the application of the principle that men and women should receive equal pay for equal work.'

3 The third paragraph of Article 119 of the Treaty specifies that:

`Equal pay without discrimination based on sex means:

(a)that pay for the same work at piece rates shall be calculated on the basis of the same unit of measurement;

(b)that pay for work at time rates shall be the same for the same job.'

4 Article 1 of Directive 75/117 reads as follows:

`The principle of equal pay for men and women outlined in Article 119 of the Treaty, hereinafter called "principle of equal pay", means, for the same work or for work to which equal value is attributed, the elimination of all discrimination on grounds of sex with regard to all aspects and conditions of remuneration.

In particular, where a job classification system is used for determining pay, it must be based on the same criteria for both men and women and so drawn up as to exclude any discrimination on grounds of sex.'

5 Article 4 of the Directive provides:

`Member States shall take the necessary measures to ensure that provisions appearing in collective agreements, wage scales, wage agreements or individual contracts of employment which are contrary to the principle of equal pay shall be, or may be, declared null and void or may be amended.'

B - National provisions

6 It appears from the order for reference and from the observations submitted to the Court that in Austria, under the Allgemeines Sozialversicherungsgesetz (General Law on Social Security, `the ASVG'), the provisions concerning the conditions of employment and the pension rights of the staff of social insurance institutions are set out in staff regulations embodied in collective agreements drawn up by the National Association of Austrian Social Insurance Institutions. Under Paragraph 460 of the ASVG, any derogation from those provisions requires the written consent of the National Association.

Dienstordnung A (Staff Regulation A) applies to employees of social insurance institutions who are engaged in administration, patient care or dental technicians' duties. Paragraph 37 of Staff Regulation A, which provides for the classification of administrative employees according to the salary bracket to which they belong, includes in `senior personnel', Salary Bracket F, Staff Category I, `psychologists authorised to practise psychology on a self-employed basis (general psychologists and clinical psychologists)'. Other psychologists are classified as `middle-ranking personnel', Salary Bracket E, Staff Category III. Paragraph 38 of Staff Regulation A, which relates to patient care staff, classifies psychotherapists in Salary Bracket C, Staff Category II.

Dienstordnung B (Staff Regulation B) applies to all doctors and dentists employed by the social insurance institutions (Paragraph 1(1)). Doctors who work in hospital establishments and are authorised to practise as specialists on a self-employed basis are classified in Salary Bracket B III.

For comparison, in 1995 the basic pay of doctors in Salary Bracket B III was between ATS 42 194 and ATS 73 457, depending on their length of service, whereas during the same period that of administrative employees in Salary Bracket F I was between ATS 24 796 and ATS 51 996, also depending on length of service.

Moreover, the working week of doctors employed by the social insurance institution which is the defendant in the main proceedings is 36 hours, whereas that of other employees is 40 hours.

The abovementioned collective agreements include detailed conditions of service and provide inter alia that subject to certain conditions, in particular completion of 10 years' service, employees are protected against arbitrary dismissal.

7 The Psychotherapiegesetz (Law on Psychotherapy) (2) defines psychotherapy (3) and lays down the conditions for practising as a psychotherapist. (4)

The Ärztegesetz (Law on Medical Practice) (5) defines the activities of the medical profession (6) and lays down the requirements for practising as a doctor. (7)

Finally, the Psychologengesetz (Law on Psychologists) (8) defines the work of psychologists (9) and lays down the requirements for practising as a psychologist. (10)

III - Facts

8 The dispute between the parties to the main proceedings, the Angestelltenbetriebsrat der Wiener Gebietskrankenkasse (Staff Committee, Vienna Area Health Fund, `the Staff Committee') and the Wiener Gebietskrankenkasse (Vienna Area Health Fund, `the Health Fund'), concerns the pay of psychologists qualified as doctors who work for the Health Fund as psychotherapists.

9 It appears from the observations (not disputed) of the Health Fund, the respondent in the main proceedings, that the said Fund is one of 28 Austrian social insurance institutions. To carry out its task, which is to implement the statutory sickness insurance scheme for employees under private-law employment contracts in the Bundesland Wien, the Health Fund uses, inter alia, numerous out-patient clinics. Many of these offer insured persons the direct psychotherapy services of employees of the insurance funds. The out-patient clinics are also used by other social insurance organisations.

10 The Health Fund employs three categories of psychotherapists: (a) qualified doctors who have completed their general or specialist training, which is obviously of a wider scope than training in psychotherapy, (b) graduate psychologists who are qualified to practise on a self-employed basis (general psychologists and clinical psychologists) and who obviously also meet the requirements laid down by the Law on Psychotherapy for practising as psychotherapists, and (c) psychotherapists who are neither doctors nor psychologists, but have completed the training laid down by the Law on Psychotherapy. It appears that in practice, the Health Fund concludes different collective employment agreements with the members of each of the above-mentioned categories, depending on their `more thorough training', whereas the order for reference indicates that all the employees concerned perform the same work of psychotherapy.

11 The Staff Committee which is the appellant in the main proceedings applied to the Arbeits- und Sozialgericht Wien (Labour and Social Security Court, Vienna) for a ruling that Staff Regulation B applies to the employment relationship between the Insurance Fund and the psychotherapists employed by it on 1 December 1994 who have a degree in psychology and that such psychotherapists should be assigned to Category B III of Staff Regulation B and be paid accordingly because, firstly, the psychotherapists in question carry out essentially the same duties as doctors who are psychotherapists, those duties being invoiced by the Health Fund at the same rate, and, secondly, it is mainly women who are affected by the fact, for which there is no objective justification, that the pay for psychotherapists who have degrees in psychology is lower.

12 The Health Fund claimed that the application should be dismissed, primarily on the ground that the different classification of these two categories is due to the fact that their training and qualifications are different and that graduate psychologists practising psychotherapy - and who are not doctors - receive in any case a higher salary than that of any graduates who do not perform executive functions, and their classification in the category sought by the Staff Committee would ultimately be equivalent to treating them as specialist doctors. The Health Fund added that the psychologists in question are not a minority in relation to doctors practising psychotherapy and that they are not engaged in what is typically a women's profession. The fact that in the Health Fund there are more women than men among psychologists working as psychotherapists is mere chance, and generally there are more women than men working as doctors in the out-patient clinics of the Health Fund.

13 The action was dismissed at first instance on the ground that the 1979 Gleichbehandlungsgesetz (Law on Equal Treatment) does not apply to all forms of differential treatment within occupational groups, but only provides for the equal treatment of men and women at work. The court also found that the Health Fund made no distinction on the basis of sex when recruiting doctors and psychologists, the disproportion between men and women being purely a matter of chance, and therefore there was no breach of the principle of equal treatment or of Article 119 of the Treaty and the Community directives based on that article. Finally, the court took the view that the different treatment of doctors in relation to graduate psychologists was attributable in any event to the functions of the former, because doctors who were appointed as specialists must, in emergencies, carry out other medical tasks, which psychologists are neither able nor entitled to do.

14 The Staff Committee appealed against that judgment to the Oberlandesgericht Wien, which established that the parties agreed on the following facts:

- the Health Fund employs a total of 248 doctors, of whom 135 are women;

- in the Ambulatorium Mariahilf out-patient clinic the Health Fund employs as psychotherapists 6 psychologists, of whom 5 are women, and 6 doctors, of whom 1 is a woman;

- the total number of persons employed as psychotherapists by the social insurance institutions is 34, of whom 10 are doctors and 24 are graduate psychologists. Of the 24 psychologists working as psychotherapists, 18 are women whereas, of the doctors working as psychotherapists, only 2 are women;

- for the whole of Austria on 9 April 1996 there were 1 425 men and 2 338 women registered as psychologists trained in psychotherapy. (11)

IV - The questions referred to the Court

15 The Oberlandesgericht Wien took the view that the decision in the case depended on the interpretation of certain provisions of Community law and decided, by order of 5 May 1997, to refer the following questions to the Court:

`(1) Do the terms "the same work" and "the same job" apply, for the purposes of Article 119 of the EC Treaty or Directive 75/117/EEC, where the same tasks are performed over a considerable length of time (several salary periods) by persons the basis of whose qualification to exercise their profession is different?

(2) Is it material, in deciding whether there is discrimination for the purposes of Article 119 of the EC Treaty or Directive 75/117/EEC, that:

(a)pay is fixed solely by the parties to an employment contract who are at liberty to incorporate therein the terms of collective agreements, or that

(b)minimum rates of pay are fixed for all employees in a given sector by general rules (collective agreements), or that

(c)pay is governed definitively by mandatory collective agreements?

(3) Where a collective agreement specifies, in definitive rules relating to remuneration, different levels of pay for the same work or work of equal value depending on professional qualifications, must reference be made, when selecting groups for comparison in determining whether a measure gives rise to discrimination, to

(a)the persons actually employed in the undertaking, or

(b)the employees working in the field covered by the collective agreement, or

(c)all those who are qualified to pursue the occupation in question?

(4) In such a case (Questions 2 and 3), must account be taken of the proportion of men to women in the disadvantaged group only, or in both groups?

(5) Where the tasks under consideration which are the same in both groups are only some of the tasks covered by the professional qualifications in question, must account be taken of

(a)all persons employed in the relevant context (undertakings, collective agreements - see Question 3) who have the professional qualifications in question (all specialist doctors and all psychologists), or

(b)all persons actually entitled to perform the duties in question (e.g. doctors with a specialist qualification in psychiatry), or

(c)only those who actually perform such duties?

(6) Where staff perform the same duties in an undertaking, may different training be regarded as a factor justifying lower pay? Is a broader professional qualification to be regarded as an objective factor justifying different pay, regardless of the duties actually performed in the undertaking?

Is the decisive factor therefore

(a)whether the better-paid group of employees may also be called upon to perform other tasks within the undertaking, or

(b)must it be shown that they were in fact called upon to perform other tasks?

In this connection, must account be taken of the fact that the applicable rules of the collective agreements include protection against unfair dismissal?

(7) Does it follow from Article 222 of the EC Treaty, or the application by analogy of Article 174 thereof, that any right to pay under another collective agreement (between the same parties) which may be inferred from Article 119 of the EC Treaty or Directive 75/117/EEC only arises when the Court of Justice rules that such a right exists?'

V - Replies to the questions referred to the Court

A - Preliminary observations

16 It is clear from the order for reference and the parties' observations that the Health Fund is a public-law body. The fact that it is not stated whether the employment relationship of the doctors and psychologists employed as psychotherapists by the Health Fund is governed by private law or public law does not preclude the application of the principle that men and women should receive equal pay for equal work, laid down by Article 119 of the Treaty. (12)

17 It is also common ground that there is an actual difference in the pay of doctors and psychologists who are employed as psychotherapists by the Health Fund and that it is more than superficial. If the difference observed were only superficial, it would be unnecessary to examine the present case further and neither the statistical data relating to the groups concerned nor the question of whether the difference in pay might be objectively justified need be considered. (13)

18 Furthermore, it is clear that in the present case, if there is discrimination based on sex in the matter of pay, such discrimination can only be indirect in so far as the rules governing the status of the two groups of workers, that is to say, doctors and psychologists who practise psychotherapy, in no way make sex a criterion for differentiation in pay. (14) Consequently, if there is a question concerning the application of the principle of equal treatment laid down in Article 119 of the Treaty, the national court must apply the principles developed by the Court for establishing the existence of indirect discrimination based on sex in the matter of pay. (15)

19 Finally, certain factual aspects of the present case seem to call for additional observations. To establish the status of the two groups of workers concerned, it is necessary to ascertain the exact terms of the employment relationship of each group, the criteria according to which doctors are chosen to work as psychotherapists, how that choice is related to their specialisation, if any, and whether, once chosen, they are employed only as psychotherapists or are assigned to other duties at the same time. It is also necessary to explain and establish, as the court of first instance in the main proceedings pointed out, why a body which represents the interests of all employees (the Staff Committee) should challenge an agreement concluded by other organisations representing the interests of the parties to the action.

However, it must be observed that the clarification required is a matter for the national court because it has better knowledge of the situation in the Member State concerned. In addition, the questions referred to the Court are of a general nature and replies can be given without resolving the factual issues set out above, which must be determined by the national court. (16)

B - The first question

20 In the first question, the Oberlandesgericht Wien asks whether the fact that the same duties are involved is sufficient for finding that there is `equal work' for the purpose of the first paragraph of Article 119 of the Treaty or `the same job' for the purpose of subparagraph (b) of the third paragraph of the same article, even where those duties are carried out on the basis of different professional qualifications, arising from different professional training, and where one of the groups of employees concerned is qualified, by virtue of the professional qualification required, to perform other duties in a wider field.

21 It must be observed that, although the importance of the term was originally underestimated, (17) the existence of `equal work' within the meaning of the first paragraph of Article 119 of the Treaty is of fundamental importance for the application of the principle of equal pay for men and women. (18) On the basis of the principle of proportional equality, discrimination contrary to the principle of equality arises through the application of different rules to comparable situations or the application of the same rule to different situations. (19) The principle of equal pay, which is a specific expression of the general principle of non-discrimination, presupposes that the men and women to whom it applies are in identical, (20) or at least similar, (21) situations. Obviously, the element of `same work' is a fundamental condition not only for regarding two groups of workers as comparable and their respective situations as similar, (22) but also for sex discrimination between those groups to exist. (23) For this reason, furthermore, in accordance with the principle that the person alleging facts in support of a claim must adduce proof of such facts, it is essential for a person pleading such discrimination to prove that the work in question is the same. (24)

22 To the semantic content of `equal work' for the purpose of the first paragraph of Article 119 of the Treaty must be added the idea of `work to which equal value is attributed' for the purpose of Article 1 of Directive 75/117. (25) The term `work of equal value' is wider than `similar work' or `equal work', so that it does not necessarily have to be identical or similar work. Different work may be regarded as comparable if it is found to have equal value, as in the case of work which is not mixed, that is to say, work which cannot be carried out by both sexes.

23 However, it must be noted that the Court has held that `Article 1 of Council Directive 75/117, which is principally designed to facilitate the practical application of the principle of equal pay outlined in Article 119 of the Treaty, in no way alters the content or scope of that principle as defined in the Treaty'. (26) In other words, the Court considered that the idea of `work of equal value', in the sense mentioned above, was and is included in the term `equal work' in Article 119 of the Treaty. The approach taken by Directive 75/117 to the interpretation of that article of the Treaty merely confirmed that the term `equal work' should be construed broadly. (27)

24 Even if the definition of `work of equal value' is in principle wider than that of `same work' and if the fundamental semantic difference which they imply is highlighted in the case of work which is dissimilar but of equal value, it must be admitted that the opposite situation may also arise: logically it is possible for similar work to exist which nevertheless has a different value. This could be the case if `same work' is construed as meaning the performance of `the same duty'. Consequently it is possible for two workers to carry out the same duty, but for the work they do to be of different value either because it is done under different conditions, or because the workers concerned have different experience or different skills. (28) In this connection I should like to stress that the concept of the value of work should not refer only to its financial value, but also its qualitative value. Even if the financial value usually accords with the qualitative value, in certain cases the cost fixed for work of a different qualitative level may be the same for reasons of expediency, for example, on grounds of social policy.

25 When, however, when is work `equal work' or `work of equal value' for the purpose of the first paragraph of Article 119 of the Treaty?

26 The third paragraph of that article seeks to define the term `equal work' used in the first paragraph, making a distinction between work at piece rates (indent (a)) and work at time rates (indent (b)) (29) In the first case, the Treaty repeats the term `equal work' used in the first paragraph of Article 119. In this case, however, the use of the term is influenced by the nature of the work at piece rates, when the worker is paid entirely or partly according to his or her productivity. (30) On the other hand, under indent (b) of the third paragraph of Article 119, the criterion for comparing the work done where pay is according to time rates is `the job', and not individual output.

`The same job' is therefore a specific form of the concept of `equal work' in the first paragraph of Article 119 of the Treaty in the context of work at time rates, and could be defined in each case on the basis of an objective formal classification of jobs in a service or an undertaking or a sector of activity. In that case, therefore, the definition of `equal work' must be based on objective criteria permitting a structural assessment of jobs and must not depend on whether the financial reward matches the individual result of the work in question. Most of these criteria are indicated by the classification of jobs in public administration and relate, in particular, to the conditions for carrying out the jobs in question, the sector of activity of which they form part, (31) their legislative context, (32) the level of the jobs in the hierarchy, their duration, the skills and knowledge required, the level of ability of workers from a general viewpoint, their period of service, etc. (33)

A systematic construction of Article 119 of the Treaty would therefore require acceptance of the following principles. The words `equal work' in the first paragraph of that article indicate any kind of work and refer to the nature and purpose of the jobs or duties which are being compared. The same words used in indent (a) of the third paragraph refer to work at piece rates and are based on a comparison of the individual results of work. On the other hand, `the same job' in indent (b) means work at time rates and refers to the formal components of the work in question, that is to say, the conditions on which it is done. It follows from what I have said that there may be `equal work', in the sense of the same duties, without there being `the same job', because those duties are not carried out on the same conditions or are carried out by workers with different skills.

27 If Article 119 of the Treaty and Directive 75/117 are construed together, the following conclusions result. By analogy with the pair of terms `equal work' and `the same job', in addition to the term `work of equal value' there must also exist the notion of `job of equal value'. This means that two dissimilar jobs may nevertheless have the same value. By analogy with my earlier observation, (34) it must be accepted that equal work may exist, in the sense that it involves the same duties, carried out by workers holding jobs of different value. In that case, equal work will not have equal value, at least if it is assessed qualitatively. (35)

28 With regard to the criteria for `equal work' or `work of equal value - or, a fortiori, higher value', (36) the guidance offered by the existing case-law is generally limited. This is due primarily to the fact that in most cases where the Court has been asked for a ruling on the interpretation of Article 119 it has started from the principle, which was a working hypothesis for the national court itself, that the workers concerned carried out equal work or work of equal value. A characteristic example of this is Enderby, (37) which, contrary to the argument of the Staff Committee which is the appellant in the main proceedings, did not rule on the question of the equal value of the work of speech therapists and pharmacists. (38)

29 The Macarthys judgment, (39) which is one of the rare judgments in which the Court directly defines its position on the criteria for `equal work' or `work of equal value', shows that `equal work' is an entirely qualitative concept which is exclusively concerned with the nature of the work in question. This means that the only criterion for the existence of `equal work' or `work of equal value' is the actual activity of the workers. As the German Government observes in the present case, there is `equal work' where identical or similar work is carried out in different jobs. (40)

30 In the subsequent case-law the Court does not appear to have modified essentially the position it adopted in Macarthys. The object and nature of the task in question appears to be a sufficient material criterion for determining whether two groups of workers perform the same work or work of equal value. (41) In addition, the conceptual distinctions made by Article 119 of the Treaty have not hitherto been examined by the Court, which has confined itself mainly to analysing the term `equal work' (42) in the context of the first paragraph of Article 119, and does not appear to attach importance to systematic analysis of the semantic difference between `equal work' and `the same job'. (43) More specifically, the organic and formal aspects of work, which are related to the worker's physical and intellectual skills, and also the conditions of employment have been examined only as objective criteria which may justify different pay for equal work. (44)

31 In my opinion, this case-law is explained not only by the fact that in the context of preliminary rulings the Court is frequently bound by the national court's presumption as to the existence of equal work or work of the same value. The fundamental reasoning underlying the case-law is that the definition of `equal work' and the criteria for determining whether there is `equal work' or not are used by the Court to a large extent as the basis for applying the case-law relating to the burden of proof in the context of applying the principle of equal pay for men and women.

The Court has observed that the burden of proving sex discrimination as to pay lies with the person who claims that it exists, that is to say, the worker who believes himself to be the victim of such discrimination. However, it is clear from the case-law (45) that the burden of proof may be transferred to the employer where that is found necessary in order for a worker who is the victim of overt discrimination based on sex not to be deprived of the right to have the principle of equal pay for equal work applied. (46) In the case of such discrimination, that is to say, where in practice a measure affects far more persons of one sex than of the other, it is now up to the employer (or the Member State or, generally, the person who adopted the measure) to prove that there are objective reasons justifying the difference in pay. (47)

Therefore the objective factors which the employer (or, generally, the person who is required to show objective justification for the difference in question) must rely on cannot relate to sex and, as the case-law shows, they will be connected with the service or the undertaking in question, (48) the conditions of work, (49) or, finally, necessary aims of social policy of the Member State concerned, the attainment of which makes the difference in pay suitable and necessary. (50) However, if it is accepted that the conditions of work must be examined, which would be logical, in connection with the preliminary question of whether the work in question is equal work or work of the same value, the only objective criteria which, at the later stage, could justify a difference in pay, without further reference to the factors already considered to determine whether there is equal work or work of the same value, will be exclusively the criteria connected with the needs or the specific characteristics of the service or the undertaking where the work is carried out, (51) or with overriding aims of social policy of the Member State concerned. In other words, many of the factors hitherto regarded as objective, which justify different pay for men and women and the existence of which must be proved by the employer (or the Member State or, generally, the person on whom the burden of proof lies), will henceforward be discussed in relation to the preliminary question of whether the work in question is equal work or work of the same value, on which point it is the worker (normally a woman) pleading discrimination who has the burden of proof. An employer who justifiably denies that discrimination exists will therefore only have to say that the work in question is not equal work or work of the same value, by relying on one of those factors, without having to show that the particular factor is objective and is unrelated to discrimination based on sex. The mere existence of that factor will entail differentiation in the type of work and will therefore mean that there is no discrimination contrary to the principle of equal pay laid down by Article 119 of the Treaty. (52) Clearly, in this case, it may not prove possible (not, at any rate, without upsetting the case-law relating to the reversal of the burden of proof and objective justification) to verify the existence of many instances of `indirect or disguised discrimination' which are based on neutral criteria connected with conditions of work or workers' skills, but which in practice affect persons of only one sex.

32 In the light of the foregoing, and taking account of the facts set out in the order for reference of the Oberlandesgericht Wien, it appears that three different interpretations may be adopted by the Court with regard to the first question referred.

(a) On the first interpretation, the Court may remain faithful to the Macarthys judgment (53) and take the view that, where the task is the same, that is sufficient to show `equal work' or `work of the same value' or `the same job', even if that task is performed with different professional qualifications arising from different training, and where one group of workers is qualified, by virtue of the professional qualification required, to perform other tasks in a wider field. This approach has the disadvantage of restricting the concept of `equal work' to the situation where the same task is carried out and of disregarding the semantic distinctions made in Article 119 of the Treaty. However, it has the advantage that it does not undermine the case-law on the distribution of the burden of proof. If this approach is taken on the facts of the present case, it will be clear that doctors and psychologists carry out equal work and have the same job because they perform the same task, that is to say, psychotherapy, on the basis of the training required by the Law on Psychotherapy, taking account also of the periods of training laid down by that Law.

(b) On the second interpretation, it is assumed that where both groups of workers perform the same duties, but have different professional training, the two groups do not have `the same job', even if they do the same work, in the sense that they perform the same duties. Simultaneously, the difference in training entails a difference in the conditions under which the work is done, so that the work of each group does not have the same economic or qualitative value.

From the conceptual viewpoint this interpretation is more consistent than the first. However, it has two fundamental disadvantages. Firstly, it implies that persons having `the same job' cannot possibly have different levels of training and, secondly, it raises questions concerning the ambit of the case-law on the burden of proof, as I have already explained. If this interpretation is applied to the facts of the present case, it must be accepted that, as psychotherapists who are doctors and psychotherapists who are psychologists do not have the same training, the question of discrimination does not arise because the two groups are not in the same situation for the purposes of the Court's case-law and do not perform tasks of the same value. If the Court adopts this interpretation, it may refrain from replying to the subsequent questions from the national court because, in this case, the principle of equal pay for men and women laid down by Article 119 of the Treaty will not apply.

(c) If the third interpretation, the one that I propose, is followed, the criterion of training must be regarded as applicable in two ways. It may be used as an objective criterion justifying a difference in the pay for equal work or work of equal value, and also as a criterion for comparing tasks. Furthermore, this position is consistent with the Royal Copenhagen judgment, (54) where the Court, in seeking to identify the factors permitting comparison of the situation of two groups of workers, observed that training is one factor for ascertaining whether workers are in a comparable situation. (55) In taking this position, the Court stated in essence that the training factor could be used not only as a criterion justifying objectively a difference in pay for equal work or work of equal value, as it did in the Danfoss case, (56) but also as a criterion for comparing the duties carried out by two groups of workers. In addition, the abovementioned judgment expressly confirms that there are two ways in which all the factors capable of justifying objectively a difference in pay and relating to the nature of the work and the conditions in which it is carried out may be applied. In particular, the Court stated that the national court must ascertain whether, in the light of the facts relating to those criteria, equal value may be attributed to the work in question or whether any pay differentials are objective and unrelated to discrimination on grounds of sex. (57)

33 However, if the training factor is to be used meaningfully in two ways, it cannot imply the same thing in both cases. As an objective criterion in relation to different pay for equal work or work of equal value, different training cannot be identified with the different training which, if it is found to exist, leads to the conclusion that two groups of workers do not carry out the same work or work of equal value.

In the first case, the difference in training normally consists in qualifications at different levels or, generally, different levels of training. Here, the qualifications at different levels do not entail a distinction so profound that the occupation or job could be said to be different, but they may justify a difference in the pay for the same work. The case is similar to that where two workers carry out the same work, but one is presumed, by reason of his length of service, to have greater experience and greater skill in fulfilling the demands of his work. This justifies a corresponding increase in pay, but it does not mean that his work is different. (58)

However, the difference in the level of qualifications may be very great or the difference in training may be not only quantitative, but also qualitative, so that the qualification in question is different and therefore the work is neither the same nor of equal value in relation to that of another group of workers who have not received the same training. In that case, the fundamentally different training, as the Commission puts it, may imply that the work has a different nature or purpose. Therefore it may be that employees of the different groups perform duties with a different nature or purpose within the same service or undertaking, but without doing equal work or work of equal value or without having the same job or a job of the same value, because the fundamental difference in training may alter radically the value of that work and the conditions under which it is done.

34 The fundamental nature of the difference is, in any case, like all the factors used to compare the duties at issue, a question of fact which must be determined by the national court. It must, however, base its assessment on the objective nature of the fundamental difference in training. (59) In my view, a different professional qualification for each group of workers is a legitimate objective criterion for determining whether there is a fundamental difference in training, with the consequence that the person in question does completely different work or has a completely different job. The qualification, which determines the scope of the tasks which may be assigned to a worker and which is awarded by reference to criteria connected with the worker's training and skills, is an essential element of the work, as the Health Fund points out. Therefore, if two employees have different qualifications because their training was fundamentally different, it follows that the work or the job in question is different, even if the employees carry out duties which appear to be identical.

35 In the present case, according to the facts in the order for reference, the doctors who work as psychotherapists appear to continue to practise as doctors and, if necessary, to perform the associated duties, unlike the graduates in psychology, who practise only as psychotherapists. Furthermore, as the Commission and the Health Fund point out in their observations, the doctors can make use of all their knowledge in looking after patients and that is why the Health Fund concludes with them contracts governed by Staff Regulation B, irrespective of the exact duties they perform. Consequently, according to the abovementioned facts, which must ultimately be clarified by the national court if necessary, if the training of the two groups of employees is fundamentally different and that difference is confirmed by different qualifications, the two groups do not carry out equal work or have the same job. Even though their duties, considered by reference to the purpose thereof, appear to be the same, that is to say, psychotherapy, the persons concerned possess fundamentally different knowledge and experience, and therefore fundamentally different therapeutic skills, and this has a significant influence on the work they perform. The fact that the work is invoiced at the same rate by the Health Fund does not necessarily preclude that finding because there is nothing to show that the invoices are based on the quality or actual market value of the treatment provided by each group of psychotherapists, and the invoicing may be based on considerations of social policy.

36 Therefore I propose that the Court's reply to the first question be that Article 119 of the Treaty (60) must be interpreted as meaning that there is not `equal work' or `the same job' where employees have different professional qualifications because they have received fundamentally different training, and perform the same duties over a considerable length of time (several salary periods), if those employees are recruited on the basis of those qualifications and if the qualifications relate to the duties which they perform.

38 I now wish to discuss the sixth question, as it is connected with the first. The Oberlandesgericht Wien asks whether, where employees perform the same duties in an establishment, a difference in training is to be regarded as a factor justifying a difference in pay. It also asks whether, irrespective of the duties actually performed, more extensive professional qualifications are to be regarded as an objective factor justifying a difference in pay. This leads to the further question of the decisive factor in that connection: (a) is the fact that the higher-paid group of employees can be used for other duties in the establishment a decisive factor, or (b) is actual proof of performance of other duties essential? Finally, the national court asks whether account is to be taken of the fact that the relevant rules in the collective agreement provide for protection against arbitrary dismissal.

39 First of all, it must be observed that the problems raised by the sixth question arise only if both groups of employees, in this case doctors and psychologists both practising psychotherapy, carry out equal work or work of the same value, or have the same job. Bearing in mind, however, the discussion of the previous question, the sixth question is justified in only two cases: (a) if the Court adopts the first interpretation, to the effect that performing the same duties is sufficient for there to be `equal work' or `work of equal value' or `the same job'; (b) if, in accordance with the third interpretation, the Court takes the view that, even if they have different qualifications, employees of the different groups assigned to the same duties in the same establishment have not received fundamentally different training, such as to justify the conclusion that they carry out different work or have a different job. The Court has held (61) that in such cases the national court, applying the principle of proportionality if necessary, has sole jurisdiction to assess whether the reasons put forward by the employer (or any other person on whom the burden of proof lies) to justify overt discrimination in pay between two groups of male and female workers or a pay practice which applies irrespective of the worker's sex, but which in actual fact affects more women than men, are based on objective criteria for a pay differential which are unrelated to sex discrimination.

40 The Court has held that those criteria include the training factor, meaning that the best-paid training is relevant to the performance of specific tasks entrusted to the employee, and the adaptability factor, meaning that an increase in pay is justified by willingness to adapt to changing hours and places of work, where the employer shows that such adaptability is relevant to the performance of specific tasks entrusted to the employee, but not if that factor relates to the quality of the work done by the employee. (62) That was the effect of the Court's judgment, although it considered, first, that the training factor might work to the disadvantage of female employees in so far as they have less opportunity than men for further training or take less advantage of such opportunity and, secondly, that the criterion of flexibility, understood as covering the employee's adaptability to changing hours and places of work, may also work to the disadvantage of female employees who, because of household and family duties, are not as free as men to organise their working time flexibly. (63)

41 In my opinion, the criterion of flexibility accepted by the Court in Danfoss (64) must also cover the idea of adaptability to the performance, in one and the same establishment, of different duties covered by the qualification held by the person concerned. Such flexibility is by nature absolutely neutral in relation to sex or, at least, is no different from the criterion of flexibility in the sense already accepted by the Court. By comparison with the reasoning in Danfoss, it is no more difficult for women to perform other, related, duties than to adapt to different working times in so far as such other duties do not necessarily entail a change in working times.

42 Clearly, the other, related, duties covered by the employee's qualifications must also be covered by the employment relationship. In other words, the employee must be recruited on the basis of his professional qualifications, that is to say, on the basis of the ability to perform several tasks in the establishment or undertaking. In the present case, therefore, the national court must ascertain whether, as stated by the Health Fund in the course of the oral procedure, the doctors were appointed as doctors - and not merely paid as doctors - whether they occasionally practise psychotherapy in the particular establishments to which they are assigned or were appointed exclusively as doctors to practise psychotherapy. (65)

43 Finally, it must be observed that the criterion of adaptability to different tasks in the same undertaking, which is based on the provisions protecting employees against arbitrary dismissal, does not conflict with the requirement that the objective criterion of differentiation should meet a real need of that establishment or undertaking. (66) `Real need' means not only present needs, but also future needs which are bound to arise. In the present case, the establishment of employees in their post after ten years means that there will be an undoubted future need for adaptability and flexibility on the part of employees in the undertaking.

44 However, is the question whether the employee actually uses at all times all the capabilities he has by virtue of his professional qualifications decisive?

As it stands, the Court's case-law does not appear to say that it is. According to Danfoss, (67) the criteria of training and adaptability to varying hours and varying places of work must be relevant to the performance of specific tasks entrusted to the employee. (68) This does not necessarily mean that in every case all the duties permitted by professional training or by the ability to adapt must be actually performed. As also appears from the Opinion of Advocate General Lenz, the condition laid down in Danfoss simply means that training and adaptability must be objectively related to the duty in question, that is to say, they must relate to the work done. (69)

Furthermore, in Macarthys (70) the Court states that `in cases of actual discrimination falling within the scope of the direct application of Article 119, comparisons are confined to parallels which may be drawn on the basis of concrete appraisals of the work actually performed by employees of different sex within the same establishment or service'. (71) In the present case, however, the requirement that duties be actually performed does not relate to the problem raised by the main proceedings with which we are concerned here, but is connected with the Court's refusal to allow a comparison between a female employee and a `hypothetical male employee', in the sense that the former could not claim the pay which she could demand if she were a man, even though no man was doing or previously did the same work. The present case does not involve the question of the `hypothetical male employee', which is bound up with the special features of the Macarthys case, (72) but the problem of a male employee who can, by reason of his training and, above all, his professional qualifications, perform in the same establishment several duties related to that which he carries out at any given time.

In my view, therefore, not only more advanced training, but also a different professional qualification entitling the holder to perform several related duties in one and the same establishment constitute an objective criterion for a difference in pay, in conformity with the requirements laid down in Danfoss, (73) regardless of whether the employee actually exercises at all time all the options given to him by the qualifications in question.

45 That view is consistent with the principle of equality. Since two groups of employees performing the same duties are not objectively in a similar professional situation in relation to the economic needs of the establishment and to the tasks entrusted to them, it is legitimate to pay them a different salary on the basis of proportionate equality. Thus, in the present case, in so far as it has been shown that they have an obligation to give first aid under Paragraph 21 of the Law on Medical Practice and that they may prescribe medicines and, if necessary, be assigned to other areas of health care, depending on their specialisation, it is logical that the doctors' pay should differ from that of psychologists who, by reference to the diverse services provided in the field of health care by a health fund such as the respondent, appear to be able to perform tasks of only limited scope by comparison with doctors.

46 That view is also justified on grounds of expediency connected with the job market. If employees who do not possess the necessary training or the appropriate qualification to enable them to carry out different duties if the need arose in the establishment received the same pay as those with such training or qualifications, the result would be that the establishment would recruit only employees of the second group. In the present case, why should the Health Fund in the future appoint psychologists as psychotherapists if it has to pay them the same as doctors even though they cannot be used for as many purposes as a qualified doctor, and why should the Fund not recruit only doctors who can also perform the duties of psychologists? If their respective pay were made equal, the consequences would ultimately be adverse for the group of employees whose pay would initially be increased. (74)

Therefore, in so far as the sixth question calls for a reply, I propose that the Court reply that, where there is overt discrimination in the matter of pay between two categories of men and women at work or a pay practice which is applied irrespective of sex, but which in actual fact affects more women than men:

- the employer may justify the difference in pay by reference to the criterion of professional training, if he shows that it is relevant to the specific tasks entrusted to the employee;

- the employer may justify the difference in pay by reference to the criterion of professional qualifications enabling the person concerned to perform several different tasks in the establishment, if the employer shows that that criterion is relevant to the specific tasks entrusted to the employee or may meet the establishment's need to assign employees to duties which differ from those already performed and which are covered by the qualification and the employment relationship, taking account of the provisions of collective agreements protecting employees against dismissal.

48 The second question asked by the Oberlandesgericht Wien is whether, in determining whether there is discrimination for the purposes of Article 119 of the Treaty or Directive 75/117/EEC, decisive importance must be attached to the type of collective agreement governing pay (mandatory, non-mandatory or mandatory with fixed minimum levels of pay).

49 At first sight it seems that, as the Commission states in its observations, the choice between three types of collective agreement is irrelevant on the facts here, because the staff regulations were adopted in the form of collective agreements laying down a method for fixing pay, which binds the various social insurance institutions in so far as they may alter the agreements only with the written consent of the General Association of Austrian Social Insurance Institutions, which drew up the collective agreements (Paragraph 460 of the ASVG). (75)

However, even if the choice among the three types of collective agreement is a hypothetical one, the Court is asked to reply to the question whether the mandatory type of collective agreement, that is to say, the type applicable in the main proceedings, is relevant when determining whether there is discrimination. The Court must therefore also examine the other types (non-mandatory, mandatory with fixed minimum levels of pay) in so far as this helps to establish the special features of the mandatory type. Consequently, the reply to the second question does not depend on whether the choice in question is described as hypothetical.

50 It must also be observed that the wording of the second question presents certain problems. It is not quite clear whether the national court is asking whether the existence of discrimination with regard to pay, for the purpose of Article 119 of the Treaty, depends on the nature of the system of collective agreements governing the pay of two different groups, or whether it is asking if that system must be taken into account in determining the standards applicable for a finding of discrimination contrary to Article 119.

51 With regard to the connection between the existence of discrimination for the purpose of Article 119 and the type of the collective agreement, the Court's case-law, to which the order for reference refers, suggests that the reply must be in the negative. The prohibition on discrimination between men and women applies not only to the action of public authorities but also extends to all collective agreements regulating paid labour, as well as to contracts between individuals. (76) The principle of the independence of will of the parties to a collective agreement could not therefore prevent the application of the principle of equal pay laid down in Article 119, regardless of the nature of the collective agreement (mandatory, non-mandatory or mandatory with fixed minimum levels of pay).

52 Therefore, if the Court considers that the meaning of the question is as stated above, I propose that the Court reply that for a finding of discrimination for the purposes of Article 119 of the Treaty or Directive 75/117 it is not material (a) that pay is determined exclusively by the parties to the employment contract, that is to say, they may decide whether to have the terms of the collective agreements incorporated therein, (b) that mandatory minimum levels of pay for every employee in a particular field are fixed by general rules (collective agreements), or (c) that pay is mandatorily governed entirely by collective agreements.

53 On the other hand, if the Court considers, in accordance with the interpretation which I have suggested, that the question also touches on the effect of the type of collective agreement on the creation of the framework in which the discriminatory conditions are laid down, the Court will have an opportunity, in view of the facts of the main proceedings, to clarify certain points concerning the application by the national court of the principle set out in Article 119 to collective agreements, such clarification being important for the main proceedings and also having a bearing upon the replies to the other questions referred by the national court.

54 The fact that the pay of two groups is determined by collective agreements affects the framework for the application of the principle of equal pay set out in Article 119 of the Treaty. In such a case, the national court must go further than the circumstances of a particular establishment and proceed to assess the circumstances which, in the matter of employment, depend on collective agreements. Furthermore, in the case of collective agreements for a whole sector, the parties, when fixing pay, normally take account not only of the specific work, but also of other factors such as the employment market generally or locally, the economic needs of undertakings in different sectors, and the size and the role of the different trade unions in the particular sector. Therefore the pay of each group of workers must be assessed, not by reference to their actual work in the undertaking, but, as the German Government observes, by reference to the formal characteristics of the work, which may be described by the job classification system laid down by the collective employment agreement in the context of the pay policy adopted. (77) The need to go beyond the specific undertaking and refer to the pay classification system laid down by the collective employment agreement becomes even more important where the agreement provides, as in the present case, for a definitive and mandatory pay system which leaves the employer with no power to differentiate on the basis of an individual contract of employment. (78) In this case, the collective agreement is the source of the objective criteria which may justify differences in pay. The Court has observed that ` the fact that the rates of pay have been determined by collective bargaining or by negotiation at local level may be taken into account by the national court as a factor in its assessment of whether differences between the average pay of two groups of workers are due to objective factors unrelated to any discrimination on grounds of sex'. (79)

55 It could also be argued that, where the pay of two groups is covered by two different collective agreements, the groups are probably not in similar or comparable situations. Moreover, Advocate General Lenz has observed that, where pay is determined by collective agreements for each sector, the fact that employees are in two different sectors may justify the conclusion that the requirement of equal work or work of equal value is not fulfilled. (80) However, the Court has developed certain criteria concerning the manner in which a national court should assess the fact that details of pay for two groups of employees are determined by collective agreements. For example, in Enderby, (81) the Court stated that `the fact that the rates of pay at issue are decided by collective bargaining processes conducted separately for each of the two professional groups concerned, without any discriminatory effect within each group, does not preclude a finding of prima facie discrimination where the results of those processes show that two groups with the same employer and the same trade union are treated differently. If the employer could rely on the absence of discrimination within each of the collective bargaining processes taken separately as sufficient justification for the difference in pay, he could ... easily circumvent the principle of equal pay by using separate bargaining processes'. (82)

56 As the German Government has observed, it is important for the national court to establish whether the collective agreements laying down the pay scales of certain occupations were drawn up by the same parties. If different stipulations of collective pay agreements are to be found comparable for the purpose of Article 119 of the Treaty, the contracting parties must be the same and the economic sector covered by the agreements in question must be the same. In the present case, it is clear from the order for reference that all the staff regulations were adopted by the General Association of Austrian Social Insurance Institutions and cover the same economic sector, that is to say, social insurance institutions, and during the oral procedure the parties to the main proceedings both admitted that the agreements in question had been drawn up by the same parties. In any case, however, it is still incumbent on the national court, which has the best information on the legal context and the facts of the case before it, to determine whether the abovementioned requirements are fulfilled.

57 Therefore I propose that the Court give the following reply to the second question: the prohibition on discrimination between men and women applies not only to public authorities but also extends to all collective agreements regulating paid labour, as well as to contracts between individuals. However, the fact that the rates of pay of two groups of employees were the result of collective bargaining processes may be taken into account by the national court as a factor in assessing how far differences in pay are due to objective factors unrelated to discrimination on grounds of sex. Likewise, the fact that each group is covered by a different collective agreement may be taken into account by the national court as a factor in assessing how far the two groups may be regarded as being in similar situations. Finally, the national court may take particular account of whether collective agreements relate to the same economic sector, whether they were concluded by the same parties and whether they lay down a definitive and mandatory system for determining rates of pay.

58 The third question referred by the national court concerns the composition of the groups to be compared (the persons actually employed in the undertaking, the employees covered by the collective agreement, all the persons with the qualifications in question) for the purpose of determining whether there is discrimination, where a collective agreement provides that pay for the same work or for work of equal value is to vary according to the employee's professional qualifications. The fourth question asks whether the relevant proportion of men to women is that in the disadvantaged group only, or that in both groups. Finally, still in the context of the problems touched upon by the third question, the fifth question asks about the exact groups to be compared in each context, where the same duties which both groups perform represent only some of the duties covered by the relevant professional qualification. I think it will be expedient to consider these questions together as they all relate to the context and the composition of the groups to be compared for the purpose of determining whether there is indirect discrimination.

59 Let me begin with a few general observations. In the case of indirect discrimination, the Court's case-law shows that the context and the composition of the groups to be compared for the purpose of determining whether there is indirect discrimination not only may, but also must, be subject to certain conditions. (83) This is reasonable because, if the comparison is to be relevant, it must involve representative groups of workers as it would be unacceptable simply to endorse a choice which is deliberately designed to support the argument of one of the parties. In any case, clearly, whether the comparison is relevant is a question of fact and within the exclusive competence of the national courts, (84) which, of course, does not prevent the Court of Justice from laying down certain criteria to guide the national courts in their assessment.

60 The first criterion regarding the composition of the groups to be compared can only be the context in which the group is formed. The Court's case-law shows that there is no general rule, but the context is determined according to the circumstances of each case. As the German Government points out in its observations, the decisive factor is the scope of the rules in question which are suspected of giving rise to discrimination for the purpose of Article 119 of the Treaty. Furthermore, the principle of equal pay laid down by Article 119 is clearly aimed at the body responsible for adopting the rules in question. Consequently, if they appear in a law, the groups to be compared must consist of all the employees whose pay is governed by that law. Similarly, if the rules are the result of a decision or a practice of a given employer, the groups to be compared must consist of all the employees of the employer in question.

61 Regarding the context of the groups to be compared for determining whether discrimination exists, it follows from what has already been said that the conclusions reached concerning the second question are of decisive importance, having regard to the facts set out in the order for reference. The fact that, firstly, the pay scales of the two groups, that is to say, doctors and psychologists, are determined by collective employment agreements and, secondly, those agreements, which are concluded by General Association of Austrian Social Insurance Institutions, are binding on the various social insurance institutions, on the basis of the conditions laid down in Paragraph 460 of the ASVG, and apply to those institutions as internal staff regulations, indicates that the groups of workers to be compared must be formed in the context of the collective agreements in question, not in the context of the institution concerned. (85) As the Commission observes, if account were taken of all the employees having the qualifications in question, the groups to be compared would be so heterogeneous that it would be pointless to compare them. Moreover, at that level the problem of discrimination would not arise.

62 Therefore I propose that the Court reply as follows to the third question: where a collective agreement provides that pay for the same work or for work of equal value is to vary according to the employees' professional qualifications, which is to serve as the criterion for forming the groups to be compared for the purpose of determining whether a particular measure gives rise to discrimination, the employees to be taken into account are those covered by the collective agreement.

63 With regard to the fourth question, the Court's case-law appears at first sight to have reached different conclusions depending on the particular case, as the national court points out in the order for reference. For example, in Jenkins, (86) the Court merely examined the category suffering discrimination, i.e. part-time workers, (87) whereas in Enderby (88) it compared two groups of workers, speech therapists and pharmacists, one consisting almost exclusively of women and the other almost exclusively of men. (89)

In my opinion, these differences in the case-law are due to the fact that, depending on the circumstances of each case, the Court, in order to satisfy itself of the existence of indirect discrimination, not only examines statistical data relating to the groups compared, which by nature are liable to fluctuate and may in general be unreliable, but also requires its conclusions to be supported on grounds which are as objective as possible. (90) More particularly, in Jenkins, the Court found that discrimination against women was shown by the fact that the ostensibly objective justification for different pay because of the number of working hours could entail such discrimination because it was difficult for women, by reason of household and family duties, to work on a full-time basis. (91) This finding, which reflects normal experience, is the objective basis of the judgment concerning indirect discrimination. According to the reasoning in that judgment, where an objective criterion of that kind exists, it may not be necessary to ascertain the proportion of men and women in the two groups of employees. In such a case, discrimination, even if it has the formal characteristics of indirect discrimination, is in essence much more likely to be regarded as direct discrimination because membership of the occupational group in question is determined at the outset on the basis of sex. Therefore it is sufficient for the national court to find that discrimination exists in an occupation which is, by nature, a `female occupation', so that it is unnecessary to compare the two groups. (92) In contrast, where indirect discrimination cannot be proved by the nature of the professional group, as in Enderby, the `female' or `male' nature of the occupation will be shown by a statistical analysis of one of the groups of employees in question, and that analysis must always be compared with that of the other group. (93)

Therefore, unlike the national court which, in the order for reference, states that, if one group of workers can be regarded as a typical case from the viewpoint of working conditions (as in the case of full-time workers), in principle it is sufficient to examine the proportion of men and women in the group suffering discrimination, whereas in all other cases the two groups must be compared, I think that, according to the Court's case-law, it is sufficient to examine the proportion of men and women in the group suffering discrimination only where an objective factor, normally the fact that membership of the group is determined at the outset on the basis of sex, indicates that, by reason of its nature, that group corresponds to a purely female or male occupation.

64 In any case, even where the Court appears to be satisfied with examining the group suffering discrimination, a comparative analysis of the data relating to the two groups cannot be ruled out.

65 The need for a comparative analysis of the proportion of men and women in the two groups arises, first and foremost, from the fact that, in order to find that a provision is `in principle' contrary to the principle of equal treatment, the Court requires the percentage of women in the group suffering discrimination to be `much higher' than the percentage of men and/or requires the percentage of women in the favoured group to be `much lower' than the percentage of men. (94) As I stressed in my Opinion in Seymour-Smith and Perez, (95) to ascertain whether there is a `significant difference' in the percentages in a group, the composition of the other group must also be taken into account. The other group must show a contrary trend, or identical percentages, or the same trend but much less marked than in the first group. If the percentage difference is the same or similar in both groups, the workers in the two groups are being treated in the same way and not unequally. (96)

66 Secondly, as the existence of indirect discrimination is a rebuttable presumption, the presence of a higher percentage of women not only in the group suffering discrimination, but also in the favoured group, is logically material evidence that the difference in question is objective and unrelated to discrimination on grounds of sex, which is a factor upon which the employer can rely. (97) Obviously, it could be objected that in that case it is incumbent on the employer, not the employee, to prove the proportion of men and women in the second group. Therefore the decision whether, in each case, the proportion of men and women in both groups must be analysed may depend on the burden of proof of discrimination. However, this observation does not appear to run counter to the need in principle to compare the proportion of men and women in both groups of workers. As I have already said, where the Court has merely examined the data on the group suffering discrimination, there was objective proof of indirect discrimination, based on common experience, and such proof generally confirms the original certainty concerning the data drawn from statistical analysis of the two groups. Comparative analysis of the proportion of men and women in the two groups is the rule, which applies in all cases of doubt, whereas examination of only one group is the exception. Furthermore, comparative analysis of the proportion of men and women in the two groups accords with the fundamental principle governing evidence of discrimination, based on comparing representative groups of workers, a principle which makes the choice of only one group unacceptable because it may be thought that it was made for the purposes of the case, that is to say, to support the arguments of one of the parties.

67 Therefore, irrespective of the existence of objective criteria which may lead to the conclusion that an occupation or a criterion is by nature `female', it is appropriate for the national court to carry out a comparative analysis of the proportion of men and women in each group. In the present case, I think that, having regard to the foregoing observations, the question of choice does not arise. The profession of psychologist-psychotherapist cannot, on the basis of the conditions for practising it, be described as purely female by nature, while the criterion of psychologist's or doctor's qualifications may in principle be regarded as objectively neutral as regards sex, in so far as both professional groups require a high level of professional training, so that the differences between them cannot be assumed to arise from the fact that women have had less opportunity than men for training or have taken less advantage of such opportunity. (98)

68 I think that the Court should reply to the fourth question as follows: as regards proof of indirect discrimination in a situation such as that described by the second and third questions, the proportion of men and women in both groups, that is to say, the group suffering discrimination and the favoured group, is always decisive.

69 After ascertaining the proper context for comparing the groups of workers, the national court must delimit those groups carefully, which is the problem raised by the fifth question referred by the national court. The Court of Justice has to a large extent laid down the criteria for such delimitation, so that the choice and the composition of the groups is not the result of chance and the groups can be compared. Therefore, according to the Royal Copenhagen judgment, (99) which summarises the Court's prior case-law, `consideration of whether the principle of equal pay has been observed requires a comparison between the pay of workers of different sexes for the same work or for work to which equal value is attributed. Where such a comparison involves the average pay of two groups of workers paid by the piece, it must in order to be relevant encompass groups each comprising all the workers who, taking account of a set of factors such as the nature of the work, the training requirements and the working conditions, can be considered to be in a comparable situation. The comparison must moreover cover a relatively large number of workers (100) in order to ensure that the differences found are not due to purely fortuitous or short-term factors ... It is for the national court to make the necessary assessments of the facts of the main proceedings in the light of the abovementioned criteria'. (101) It seems therefore that the abovementioned criteria apply generally. In spite of the wording of the judgment, there appears to be no reason why they should apply only to piece workers.

70 In the present case the groups to be compared must therefore be formed on the basis of the abovementioned criteria in the context of general collective agreements, as I indicated in my reply to the third question. Applying the criterion of difference in pay rates which is introduced by the collective agreements, that is to say, by reference to the criterion of qualifications, it seems that the two groups in question are, firstly, psychologists qualified to practise as such on a self-employed basis and who work in all Austrian social insurance organisations and, secondly, doctors qualified to practise as such on a self-employed basis and who work as medical specialists in all Austrian social insurance organisations. Therefore in the context of those two groups, for which the difference in pay is the result of collective agreements, it is necessary to analyse the percentages of men and women to ascertain whether there is discrimination in relation to pay. (102)

71 However, the problem raised by the fifth question is not so simple as it appears at first sight because it must not be forgotten that, for the fifth question to arise, the reply to the first must be in the affirmative.

Normally, the work of the groups to be compared must be the same or work of equal value. (103) Consequently, in order for the national court to determine in more detail the groups to be compared, in principle it must be shown that they perform the same work or work of equal value. However, as I said with regard to the first question, because the qualifications involved in the present case are different, it cannot be said that the work is the same or of equal value unless the practice of psychotherapy is regarded as the same task. Therefore, even if it is not mentioned in the collective agreements, the criterion of performing the same task must be combined with that of holding a given qualification, so that the groups concerned are in similar situations. It follows that the two groups to be compared must, first, be psychologists qualified to practise as such on a self-employed basis and who work as psychotherapists in all Austrian social insurance organisations and, secondly, doctors qualified to practise as such on a self-employed basis as specialists and who work as psychotherapists in all Austrian social insurance organisations.

However, a comparison in the context of the abovementioned groups in order to ascertain whether there is a problem of discrimination in relation to pay would not be adequate. With regard to the rates of pay of the two groups of workers, doctor-psychotherapists and psychologist-psychotherapists, the groups which are compared in terms of numbers are, as I mentioned previously, determined in the context of collective agreements, which lay down mandatory rates of pay. If, nevertheless, it is necessary to take into account the sub-groups of doctors and psychologists practising psychotherapy in all Austrian social insurance organisations, the purpose of this is only to ascertain whether their work is the same or of equal value. If, therefore, in that context there were statistics which might indicate a problem of discrimination against women, but no similar statistics in the context of collective agreements, (104) in practice that discrimination would not relate to pay but to other aspects of the work in question. In particular, there might be discrimination on grounds of sex with regard to the choice, among doctors and/or psychologists, of persons to practise psychotherapy.

However, on this point it must be stressed that it is precisely this aspect of the main proceedings, that is to say, the fact that whether the work is the same or of equal value can be assessed only in a context other than that in which the groups of employees are compared in order to ascertain whether there is a problem of discrimination in pay, which proves two things. In the first place, it shows how difficult it is to find without hesitation that two groups of employees do the same work or work of equal value - or do the same job - where the two groups happen by chance to perform the same tasks, without disregarding any aspect of their different professional status. Secondly, it shows in this particular case the complex conclusions to which we are bound to be led if we accept that idea.

72 Subject to this last observation, I propose that the Court, if it deems it appropriate to do so, reply to the fifth question as follows: where the duties in question performed by the two professional groups are comprised in the duties covered by professional qualifications, it is necessary to take account of all the employees possessing a qualification of that type who are within the scope of the collective employment agreement.

F - The seventh question

73 The seventh question seeks in essence a ruling on whether, where a given collective agreement or staff regulations entails discrimination on grounds of sex, contrary to Community law, that discrimination will be terminated only on the date of a finding to that effect by the Court of Justice or whether it will be terminated with retrospective effect.

74 The Court has consistently held that the interpretation which, in the exercise of the jurisdiction conferred on it by Article 177 of the Treaty, the Court gives to a rule of Community law clarifies and defines where necessary the meaning and scope of that rule as it must be or ought to have been understood and applied from the time of its coming into force. It follows that the rule as thus interpreted may, and must, be applied even to legal relationships arising and established before the judgment ruling on the request for interpretation, provided, of course, that in other respects the conditions enabling an action relating to the application of that rule to be brought before the courts having jurisdiction are satisfied. (105)

75 Exceptions to the abovementioned rule arise only in very exceptional cases where the Court finds that observance of the rule would run counter to the principle of legal certainty which is inherent in the Community legal order. When considering whether that principle imposes a limitation on the effect ratione temporis of a judgment given on a reference for interpretation, the Court determines whether (a) there is a risk of serious economic repercussions owing in particular to the large number of legal relationships entered into in good faith on the basis of rules considered to be validly in force, and (b) individuals and national authorities have been prompted to adopt practices which did not comply with Community law by reason of objective, significant uncertainty regarding the implications of the Community provisions interpreted by the Court; if it is found that the conduct of other Member States or the Commission gave rise to or reinforced that uncertainty, that will be particularly relevant. (106)

The Court has consistently held that such a restriction may be allowed only in the actual judgment ruling upon the interpretation sought. (107) However, that restriction does not apply to persons who brought legal proceedings or submitted an equivalent claim before that date. (108)

As the Court has held, in determining whether or not to limit the temporal effect of a judgment it is necessary to bear in mind that although the practical consequences of any judicial decision must be weighed carefully, the Court cannot go so far as to diminish the objectivity of the law and compromise its future application on the ground of the possible repercussions which might result, as regards the past, from a judicial decision. (109)

76 On the basis of the observations in the order for reference, the seventh question does not arise if, in reply to the first question, the Court finds that the work in question is `the same work' or `work of equal value'. The seventh question arises only if, in reply to the sixth question, the Court finds that, where the same tasks are performed, the different levels of training and the different qualifications do not in themselves constitute differences justifying discrimination within the meaning of Article 119 of the Treaty. Consequently the conditions for applying the abovementioned case-law must be examined only in the latter case.

77 The direct application of Article 119 has been a settled principle of Community law since the judgments in Defrenne (110) and Jenkins. (111) Since 1 July 1995, the date of the accession of the Austrian Republic to the European Union, employers and employees in Austria cannot have failed to be aware of the conditions laid down by Article 119 when negotiating collective agreements. In this connection it must be stressed that, contrary to the statements of the national court in the order for reference and of the Health Fund in its observations, the fact that legal relationships have been created in good faith on the basis of measures which were generally thought to comply with national law is irrelevant. In order for the question of legitimate expectation with regard to the effect of a judgment ratione temporis to arise, the legal relationships must be created on the basis of measures which were generally thought to comply with Community law.

On the basis of the case-law laid down in Danfoss, (112) the Member States and employers and employees would be fully entitled to believe that different levels of training, and different qualifications which are attributable to different training, constitute, in the circumstances described by the said judgment, differences objectively justifying discrimination within the meaning of Article 119 of the Treaty. Therefore, should the judgment to be given by the Court differ significantly from the principles set out in Danfoss, that really would create a risk of serious economic repercussions owing in particular to the large number of legal relationships entered into in good faith on the basis of rules considered to be validly in force under Community law, and to the fact that employers and employees have been prompted to adopt practices which do not comply with Community law by reason of objective, significant uncertainty regarding the implications of the Community provisions interpreted by the Court. What would have led to, or increased, that uncertainty is not the conduct of other Member States or of the Commission, but the case-law of the Court itself. (113)

78 Consequently I propose that the Court limit the future effect ratione temporis of its judgment for all those persons who have not hitherto brought legal proceedings or submitted equivalent claims only if it refuses to follow the Danfoss judgment and ultimately decides that, where the same tasks are performed, the different levels of training and the different professional qualifications do not in themselves constitute differences justifying discrimination within the meaning of Article 119 of the Treaty.

VI - Conclusion

79. I propose that the Court reply as follows to the questions referred to it by the Oberlandesgericht Wien:

(1) Article 119 of the EC Treaty must be interpreted as meaning that `equal work' or `the same job' does not exist where employees have different professional qualifications because they have received fundamentally different training and perform the same duties over a considerable length of time (several salary periods), if those employees are recruited on the basis of those qualifications and if the qualifications relate to the duties which they perform.

(2) If the foregoing reply is given to the first question, it is unnecessary to reply to the others.

(1) - OJ 1975 L 45, p. 19.

(2) - Bundesgesetz über die Ausübung der Psychotherapie, BGBl. 361/1990.

(3) - Psychotherapy is defined as the comprehensive, considered and structured treatment, on the basis of general and specialist training, of behavioural disorders and psychopathic conditions caused by psycho-social or psychosomatic factors, by means of scientific psychotherapeutic methods in interaction between one or more patients and one or more psychotherapists, with the aim of alleviating or eliminating existing symptoms, modifying disturbed behaviour and attitudes and promoting the patient's maturity, development and health (Paragraph 1).

(4) - To practise as a psychotherapist it is not necessary to have completed a university course, but a general training course (basic psychotherapy training module) consisting of not less than 765 hours of theory and 550 hours of practical training must have been followed. The further training necessary (specialist psychotherapy module) consists of a minimum of 300 hours of theory and 1 600 hours of practical training. In addition to possessing certain qualifications (see Paragraph 10(2)(5) to (9)), practising psychotherapists must also be registered in the register of psychotherapists (Paragraph 17). The right to practise the profession under the Psychologengesetz (Law on Psychologists) is not restricted to persons satisfying these requirements (Paragraph 24 (3)).

(5) - Bundesgesetz über die Ausübung des ärztlichen Berufes und die Standesvertretung der Ärzte, BGBl. 373/1984.

(6) - These are all activities requiring knowledge of medical science, which are applied directly to or indirectly for humans, including in particular the diagnosis and treatment of mental illness and emotional disorders (Paragraph 1(2)(1) in conjunction with Paragraph 1(2)(3) of the Law on Medical Practice). Consequently these activities include the practice of psychotherapy.

(7) - In order to practise as a medical specialist it is necessary to complete at least six years' postgraduate practical training, for which there are comprehensive regulations, and to pass the qualifying examination for specialists (Paragraph 5 of the Law on Medical Practice).

(8) - Bundesgesetz über die Führung des Berufsbezeichnung `Psychologe' oder `Psychologin' und über die Ausübung des psychologischen Berufes im Bereich des Gesundheitswesens, BGBl. 360/1990.

(9) - The practice of psychology in the field of health consists in the investigation, interpretation, modification and prognosis of human experience and behaviour by applying the knowledge and methods of the science of psychology gained through the acquisition of specialist competence within the meaning of the federal law in question.

(10) - Psychology graduates must take a specialised postgraduate course of 160 hours of theory, 1 480 hours of practical training and an additional 120 hours of training in the particular field of speciality.

(11) - According to the figures referred to by the Insurance Fund in its written observations, the 28 social insurance institutions in Austria and the National Association of Social Insurance Institutions employ 29 633 persons, of whom 18 634 (62.88%) are women. The social insurance institutions employ 1 932 doctors. 536 doctors work in out-patient clinics and 260 of these (48.51%) are women. The social insurance institutions employ a total of 50 psychologists. 14 psychologists work in out-patient clinics and 13 of these (92.86%) are women. Finally, during the written procedure the Insurance Fund stated that 50% or 51% of the doctors working for social insurance institutions are women, and this figure has not been challenged by the other parties.

(12) - The Court has in any case held that this principle applies to the public service: see Case C-1/95 Gerster [1997] ECR I-5253, paragraphs 18 and 19.

(13) - See Joined Cases C-399/92, C-409/92, C-425/92, C-34/93, C-50/93 and C-78/93 Helmig and Others [1994] ECR I-5727, paragraphs 23, 30 and 32.

(14) - For the definition of indirect discrimination, see Gerster (cited in footnote 12, paragraph 30) and the definition which now appears in Article 2(2) of Council Directive 97/80/EC of 15 December 1997 on the burden of proof in cases of discrimination based on sex (OJ 1998 L 14, p. 6).

(15) - As to those principles, see my Opinion of 14 July 1997 in Case C-167/97 Seymour-Smith and Perez, still pending, paragraph 117 et seq.

(16) - Case C-127/92 Enderby [1993] ECR I-5535, paragraph 10.

(17) - According to Advocate General Trabucchi (see his Opinion in Case 43/75 Defrenne [1976] ECR 455, in particular p. 486), `with regard to the definition of the concept of "equal work", which is in any case partly described in the third paragraph of Article 119 (which, in addition to using the term "same work" for work at piece rates, also refers to pay for work at time rates as "the same for the same job"), there is no need to exaggerate its importance in applying the article. It has been rightly observed that Article 119 "does not try to determine when men and women are doing the same work, but only to ensure that the sex of the worker is in no way taken into account in decisions on pay. Whether the work is the same or different is a question of fact to be determined in every individual case in accordance with the responsibilities assigned to each person concerned and must not be the subject of an a priori decision any more than there is an a priori decision that two men placed on the same rate of pay perform the same work" (Levi Sandri, in Commentario CEE, vol. II, p. 956).'

(18) - See the Opinion of Advocate General Lenz in Enderby (cited in footnote 16, paragraph 5).

(19) - Case C-279/93 Schumacker [1995] ECR I-225, paragraph 30; Case C-342/93 Gillespie and Others [1996] ECR I-475, paragraph 16, and Case C-100/95 Kording [1997] ECR I-5289, paragraph 14.

(20) - Case C-132/92 Roberts [1993] ECR I-5579, paragraph 17. The actual wording of this judgment indicates that the work in question must be absolutely identical. However, the wording does not accord with the more general spirit of the Court's case-law. On this point, see paragraphs 22 and 23 of this Opinion.

(21) - Case C-400/93 Royal Copenhagen [1995] ECR I-1275, paragraph 33.

(22) - See the Opinion of Advocate General Lenz in Case 109/88 Danfoss [1989] ECR 3199, paragraph 40, and the Royal Copenhagen judgment, cited in footnote 21, paragraphs 32 and 33.

(23) - Royal Copenhagen, cited in footnote 21, paragraph 40.

(24) - Enderby, cited in footnote 16, paragraphs 13 to 19, and the Opinion of Advocate General Lenz in that case, paragraphs 15 to 39, and also his Opinion in Roberts, cited in footnote 20, paragraph 12 et seq.

(25) - Directive 75/117 filled the gap in Article 119 in relation to Article 2(1) of Convention 100 of the International Labour Organization (1951) and Article 4, point 3, of the European Social Charter (Turin, 18 October 1961), both of which laid down the principle of equal treatment for work of equal value.

(26) - See Case 96/80 Jenkins [1981] ECR 911, paragraph 22, and Case 192/85 Newstead [1987] ECR 4753, paragraph 20.

(27) - See the Opinion of Advocate General Capotorti in Case 129/79 Macarthys [1980] ECR 1275, to the effect that `same work' in Article 119 `should be understood ... in such a manner as to include two jobs showing a high degree of similarity, even if there is not total identity between them' (point 4, in particular p. 1296). In any case, since the Treaty of Amsterdam (Article 1, point 22), Article 119 of the EEC Treaty now refers expressly to two concepts: `1. Each Member State shall ensure that the principle of equal pay for male and female workers for equal work or work of equal value is applied.'

(28) - The difference in skills may be so great that it gives rise to the question whether the job or the work is completely different: see paragraph 33 of this Opinion.

(29) - The consequences of this distinction, which the Court has not yet examined in detail, must be assessed in the light of the fact that there are also mixed systems. See also Royal Copenhagen (cited in footnote 21), which deals with this distinction.

(30) - His pay may include a fixed portion, but the main part is variable and is paid by reference to the number of articles produced. Thus his actual earnings are determined, at least in part, by his output, which is assessed on a wholly individual basis (see the Opinion of Advocate General Léger in Royal Copenhagen, cited in footnote 21, paragraph 60). In the case of piece-work, therefore, a comparison of output involves comparing the individual productivity of each worker.

(31) - See paragraph 55 of this Opinion, concerning the relevance of the sector to the question whether there is equal work or work of equal value.

(32) - See paragraph 54 et seq. of this Opinion, concerning the second question referred, which raises the issue of the relevance of provisions in collective agreements in the assessment of whether the work is the same or of equal value.

(33) - On comparing managerial posts, see the Opinion of Advocate General Lenz in Enderby, cited in footnote 16, paragraph 10.

(34) - See paragraph 24 of this Opinion.

(35) - See paragraph 24 of this Opinion.

(36) - See Case 157/86 Murphy and Others [1988] ECR 673, where the Court held that Article 119 of the Treaty must be interpreted as covering the case where a worker who relies on that provision to obtain equal pay within the meaning thereof is engaged in work of higher value than that of the person with whom a comparison is to be made (paragraph 12). To adopt a contrary interpretation would render the principle of equal pay ineffective and nugatory (paragraph 10).

(37) - Cited in footnote 16.

(38) - Enderby, cited in footnote 16. As the questions of interpretation of Community law referred by the national court were not manifestly unrelated to the reality or the subject-matter of the main proceedings, the Court adopted the hypothesis of the national court that the work was of equal value. The Court observed that the validity of that hypothesis must be verified subsequently by the national court (paragraphs 10 to 12). See also Danfoss, cited in footnote 22. On this point, in Royal Copenhagen, cited in footnote 21, the Court adopted the national court's hypothesis that the work in question was of equal value, although there may have been some doubt as to the validity of that hypothesis, as Advocate General Léger indicated in his Opinion (paragraphs 14 to 17).

(39) - Cited in footnote 27.

(40) - Macarthys, cited in footnote 27, paragraph 11. As to the question of work done at different times, the Court observed that the fact that the work compared is done at different times is, in principle, irrelevant as regards discrimination which is directly covered by Article 119 (paragraph 12). However, the possibility cannot be ruled out that a difference in the pay of workers doing the same job but at different times may be explained by the operation of factors which are unconnected with discrimination on grounds of sex, for example where in the meantime economic conditions have changed or a stricter incomes policy is adopted. This is a question of fact which must be assessed by the national court.

As to the question of jobs done at different places, the Court ruled in the same judgment that it was not possible to compare the work of a man and a woman in a very general framework, taking account of entire sectors of activity; comparisons had to be confined to the limited framework of the same establishment of the employer to which the case relates. Likewise, the comparison must relate to the specific characteristics of the jobs in question, arising from the tasks actually carried out (Macarthys, cited in footnote 27, paragraphs 10, 14, 15 and 16).

As regards jobs done at different places, the Macarthys judgment (like the judgment in Case 69/80 Worringham and Humphreys [1981] ECR 767; see also the Opinion of Advocate General Warner, who took the opposite view) is no longer of particular importance because it was based on the case-law beginning with the Defrenne judgment, cited in footnote 17, concerning the direct effect of Article 119. According to those judgments, Article 119 has direct effect only in the case of `direct and overt discrimination' which may be identified solely with the aid of the criteria based on equal work and equal pay referred to by the article in question, and it is not necessary for there to be more explicit definitions of those criteria in provisions of a Community or national character. In line with that case-law restricting the direct effect of Article 119 to direct discrimination, the Court refused to consider whether discrimination might arise on the basis of a comparative study of tasks, extending to entire branches of industry, because it considered that the discrimination in question was indirect. For the same reason, the Court took the view that it was not possible to make a comparison by reference to `a hypothetical male worker' (see Macarthys, cited in footnote 27, paragraphs 14 and 15). However, the Court is no longer reluctant to extend the direct application of Article 119 to indirect discrimination. It now accepts that the issue of indirect discrimination should be left to the national court. Likewise, on the basis of Article 1(2) of Directive 75/117, the physical parameters of a comparative study of tasks may be enlarged where unequal treatment is found in a sector by virtue of a collective agreement or of legislation. See, for example, the Opinion of Advocate General VerLoren van Themaat in Case 143/83 Commission v Denmark [1985] ECR 427; Jenkins (cited in footnote 26), Case C-184/89 Nimz [1991] ECR I-297, Case 171/88 Rinner-Kühn [1989] ECR 2743 and Case C-33/89 Kowalska [1990] ECR I-2591.

(41) - Case 237/85 Rummler [1986] ECR 2101. This judgment refers mainly to objective criteria for determining the existence of discrimination in pay. The principle of equal pay to which the judgment refers primarily requires objective account to be taken of the nature of the work. The work actually carried out must be remunerated in accordance with its nature (paragraph 23). In Murphy (cited in footnote 36), the higher value attributed to the work of one of the two groups of workers compared is also based on the nature and purpose of that work.

(42) - Macarthys, cited in footnote 27, paragraph 9.

(43) - For example, in his Opinion in Jenkins (cited in footnote 26), Advocate General Warner argued that `equal work' and `the same job' were not synonymous and he concluded that full-time and part-time workers did not do the same job even though their work was `equal'. This observation was linked to the question referred, which was whether the requirement of `equal work' was fulfilled, that being a question which logically precedes any other question because if the work done by both groups of workers could not be regarded as equal work or work of the same value, there would be no grounds for applying Article 119. The Court implicitly accepted that the fact that one group worked full-time and the other part-time did not prevent that condition from being fulfilled. The Court found it sufficient that the task was the same and did not go on to consider whether `the same job' was involved (paragraphs 10 to 15).

In the only case where a clear distinction was made between work at piece rates and work at time rates and where this distinction was found decisive for assessing the consequences of a comparative study of the average pay of two groups of workers (see Royal Copenhagen, cited in footnote 21, paragraph 25), the Court did not make a systematic analytical interpretation of the conditions laid down in Article 119 of the Treaty, although it had the opportunity to do so. In laying down the criteria for comparing the work done by the two groups of workers, the Court did not rely on the conceptual distinction between `equal work' and `the same job'. It referred in succession to the substantive criterion of the nature of the work, the formal criterion of the conditions in which it is carried out, and the criterion of the training received by the workers. It appears from the wording of the judgment that the Court considered the last-mentioned criterion to be distinct from the others (paragraphs 32, 33 and 42).

(44) - With regard to the criterion of working hours (full- or part-time), see, for example, Jenkins (cited in footnote 26). For the criterion of training, with which the present case is concerned, see Danfoss (cited in footnote 22, paragraph 23).

(45) - Enderby, cited in footnote 16, paragraphs 13 to 19.

(46) - See also Article 4(1) of Directive 97/80 (cited in footnote 14).

(47) - Thus the Court has found that where a measure distinguishing between persons according to working hours affects a much greater number of persons of one or the other sex, that measure must be deemed contrary to the aims of Article 119 of the Treaty, unless the employer proves that the measure is based on objectively justified factors unrelated to any discrimination on grounds of sex (Case 170/84 Bilka [1986] ECR 1607, paragraph 31; Kowalska, cited in footnote 40, paragraph 16, and Nimz, cited in footnote 40, paragraph 15). In addition, the Court has stated that, where an undertaking applies a system of pay which is totally lacking in transparency, it is for the employer to prove that his practice in the matter of wages is not discriminatory, if a woman employee establishes, in relation to a relatively large number of employees, that the average for women is less than that for men (Danfoss, cited in footnote 22, paragraph 16). The Court has also observed that where significant statistics disclose an appreciable difference in pay between two jobs of equal value, one of which is carried out almost exclusively by women and the other predominantly by men, Article 119 of the Treaty requires the employer (or, generally, the person who must justify the distinction) to show that the difference is based on objectively justified factors unrelated to any discrimination on grounds of sex (Enderby, cited in footnote 16, paragraph 19).

(48) - Bilka, cited in footnote 47, paragraph 37.

(49) - Danfoss, cited in footnote 22, with regard to length of service, training, adaptability in general and specifically in relation to variable hours and places of work.

(50) - See, for example, Rinner-Kühn (cited in footnote 40, paragraph 14), Gerster (cited in footnote 12, paragraph 40) and Case C-457/93 Lewark [1996] ECR I-243, paragraph 36. Likewise the Court has considered as an objective criterion a policy of economic development and job creation (Case C-189/91 Kirsammer-Hack [1993] ECR I-6185, paragraph 33). In contrast, the Court found that considerations of a budgetary nature could not be objective criteria (Case C-343/92 De Weerd (née Roks) and Others [1994] ECR I-571, paragraphs 35 and 36).

(51) - However, even these criteria, in so far as they are not merely of an economic nature and in so far as they may concern the framework in which the work is carried out, are functional criteria for assessment of the work and should be examined in the context of the preliminary question of whether the work in question is equal work or work of equal value.

(52) - See paragraph 21 of this Opinion.

(53) - Cited in footnote 27.

(54) - Cited in footnote 21.

(55) - Royal Copenhagen, cited in footnote 21, paragraphs 32 and 33.

(56) - Cited in footnote 22.

(57) - Royal Copenhagen, cited in footnote 21, paragraph 42.

(58) - Danfoss, cited in footnote 22. However, as the Court observed in that case, with regard to the length of service, the conditions are not the same as for the training criterion, or adaptability to different working hours or places of work, because `length of service goes hand in hand with experience and ... experience generally enables the employee to perform his duties better' (paragraph 24).

(59) - The Court has observed that an important criterion for assessing the teaching and academic skills of a given category of university teachers or researchers is participation in an open competition or recognition by a national committee: see Case C-90/96 Petrie and Others [1997] ECR I-6527, paragraph 47 et seq.

(60) - In the reply it is sufficient to refer to Article 119 of the Treaty and it is not necessary to add a reference to Directive 75/117: see Macarthys, cited in footnote 36, paragraph 13, and paragraph 23 of this Opinion.

(61) - Enderby, cited in footnote 16, paragraphs 18, 19 and 25, and Bilka, cited in footnote 47, paragraphs 35 and 36.

(62) - Danfoss, cited in footnote 22, paragraphs 17 to 25, which states that application of the abovementioned criteria is subject to the requirement that they are not based on generalisations about certain categories of workers, but are objective, which depends on the circumstances of each case and, above all, their relationship with the nature of the work performed. Regarding this requirement, see Rinner-Kühn, cited in footnote 40, paragraph 14, and Nimz, cited in footnote 40, paragraphs 13 to 15) which, contrary to the observations of the Staff Committee, do not deny that the training criterion is legitimate and objective.

(63) - Danfoss, cited in footnote 22, paragraphs 21 and 23.

(64) - Cited in footnote 22.

(65) - On this point, it may be appropriate also to consider the other problems connected with the possibility of performing several related tasks. For example, it may be appropriate for the national court to consider whether medical ethics permit doctors practising in a specialist field (for example, psychiatrists who, in the present case, are employed as psychotherapists) to practise in another.

(66) - Bilka, cited in footnote 47, paragraph 36.

(67) - Cited in footnote 22.

(68) - See paragraph 40 of this Opinion.

(69) - See the Opinion of Advocate General Lenz, cited in footnote 22, paragraph 43.

(70) - Cited in footnote 27.

(71) - Paragraph 15.

(72) - See paragraph 40 of this Opinion.

(73) - Judgment cited in footnote 22.

(74) - These considerations should not lead to the conclusion that the work of doctors, or their training, is bound to be of a qualitatively higher level than that of psychologists. The fact that they are different does not necessarily mean that they are better. Moreover, it is well-known that, in the context of psychotherapy, psychologists use methods of diagnosis and treatment (for example, tests) with which psychiatrists, and particularly doctors who are not specialists in psychiatry, are not generally familiar. Likewise, generally speaking, psychologists can probably perform duties other than psychotherapy. In the present case, however, it is significant that, having regard to the health-care services offered by a health fund such as the respondent in the main proceedings and given the professional training which they have received, doctors appear to be employable, within that Fund, in more areas, and this clearly makes them more competitive than psychologists in the job market, as the Health Fund points out. However, it is for the national court, which is familiar with the national legislative framework and the facts of the main proceedings, to assess the extent to which doctors can be employed in more areas and their more competitive position in the job market.

(75) - For the Court's refusal to reply to hypothetical questions, see Case C-83/91 Meilicke [1992] ECR I-4871, paragraph 25.

(76) - See Article 4 of Directive 75/117, cited in paragraph 4 of this Opinion. See also Royal Copenhagen (cited in footnote 21, paragraph 45), Defrenne judgment (cited in footnote 17, paragraph 39), Kowalska (cited in footnote 40, paragraph 12), Nimz (cited in footnote 40, paragraph 11) and Enderby (cited in footnote 16, paragraph 22 et seq.). According to the judgment in Case C-262/88 Barber [1990] ECR I-1889, paragraph 32), Article 119 of the Treaty prohibits any discrimination with regard to pay as between men and women, whatever the system which gives rise to such inequality.

(77) - According to the Rummler judgment (cited in footnote 41, paragraph 15), `even where a particular criterion ... may in fact tend to favour male workers ... it must, in order to determine whether or not it is discriminatory, be considered in the context of the whole job classification system, having regard to other criteria influencing rates of pay. A system is not necessarily discriminatory simply because one of its criteria makes reference to attributes more characteristic of men'.

(78) - But see the judgment in Case 165/82 Commission v United Kingdom [1983] ECR 3431. The Court stated that binding collective agreements also have important de facto consequences for the employment relationship to which they refer, particularly in so far as they determine the rights of the workers and, in the interests of industrial harmony, lay down the limits which must be imposed on the requirements of undertakings.

(79) - See Royal Copenhagen (cited in footnote 21, paragraph 46).

(80) - See the Opinion in the Enderby case (cited in footnote 16, paragraphs 44 to 46) and in the Danfoss case (cited in footnote 22, paragraph 46).

(81) - See the judgment cited in footnote 16.

(82) - Paragraph 22.

(83) - See Enderby (cited in footnote 16, paragraph 17) and Danfoss (cited in footnote 22, paragraph 16).

(84) - See the Opinion of Advocate General Léger in Royal Copenhagen (cited in footnote 21, paragraphs 28 to 30).

(85) - However, it is important for the national court to establish whether the fund or the undertaking in question takes specific steps (practices, assignments, etc.) to determine which doctors (according to their specialisation) or psychologists will be recruited and will practise psychotherapy, so that the permanent difference in pay of men and women is essentially the result of those steps.

(86) - See the judgment cited in footnote 26.

(87) - Paragraphs 9 et seq., in particular paragraph 15. See also Kowalska, paragraph 16, and Nimz, paragraph 15, both cited in footnote 40. The Bilka judgment (cited in footnote 46, paragraph 29) examines only the favoured category, i.e. full-time workers.

(88) - See the judgment cited in footnote 16.

(89) - Paragraphs 15 to 19; see also Royal Copenhagen (cited in footnote 21, paragraph 28).

(90) - See my Opinion in Seymour-Smith and Perez, cited in footnote 15, paragraphs 113, 114 and 119. With regard to the coexistence of an examination of statistical data and of objective factors which may justify a difference in treatment, see the Helmig judgment, cited in footnote 13, paragraphs 23 to 25, which refers to Bilka, cited in footnote 47.

(91) - Paragraph 13. See Danfoss, cited in footnote 22, paragraph 21.

(92) - According to the judgment in Case C-177/88 Dekker [1990] ECR I-3941, paragraphs 12 and 17, a comparison is unnecessary where the criterion is specifically connected with sex, that is to say, where, for example, it could never refer to men (if the criterion is pregnancy), whereas it is necessary if discrimination can be justified only by a comparison of different treatment of men and women.

(93) - See the Opinion of Advocate General Lenz in Enderby (cited in footnote 16, paragraphs 30 and 31).

(94) - For the category of full-time workers, see Bilka (cited in footnote 47, paragraph 29).

(95) - Cited in footnote 15.

(96) - See footnote 50 of my Opinion in Seymour-Smith and Perez, cited in footnote 15.

(97) - See Kowalska (paragraph 16) and Nimz (paragraph 15), both cited in footnote 40.

(98) - See Danfoss (cited in footnote 22, paragraph 23).

(99) - Cited in footnote 21.

(100) - However, the question of specifying a minimum number did not arise. See the Opinion of Advocate General Léger in that case (paragraph 30).

(101) - Paragraphs 32 to 35. See also Enderby (cited in footnote 16, paragraph 17).

(102) - It should be observed that, with regard to the context of the groups in question, the necessary statistics are not given clearly in the documents in the file. It gives no exact figures concerning the percentages of men and women among all qualified psychologists or among all doctors working in all Austrian social insurance organisations. Some figures were given by the Insurance Fund in its written observations and during the oral procedure (see footnote 11 of this Opinion). However, in all cases it is for the national court to decide whether the statistics provided by the parties are sufficient to establish the existence of indirect discrimination (see Enderby, cited in footnote 16, paragraph 17).

Regarding the confusion concerning the choice of suitable statistics, however, although the Commission states that groups of workers must be compared in the context of collective agreements, during the oral procedure the Commission compared psychologists practising psychotherapy with doctors in general and, in addition, in its written observations it compared the figures which, so far as doctors are concerned, relate to all employees of the Insurance Fund and, for qualified psychologists, relate to all employees practising psychotherapy in the sector of social insurance organisations (see the Commission's written observations, paragraph 48). Consequently the conclusion to which the Commission is led, that women form the majority among both doctors and psychologists, is not persuasive because it is not founded on suitable statistical data.

At the same time, the way in which one must determine the context in which the comparison of groups of workers must be made, the question I have just dealt with, invalidates the Health Fund's remark that the fact that the main proceedings concern only one out-patients' clinic belonging to only one social insurance organisation and, within that clinic, only six doctors and six psychologists, whereas the rules on the status of employees of social insurance organisations apply to a total of 1 932 doctors, 50 psychologists and 27 651 other workers, is significant in the light of the Court's case law to the effect that, for determining whether there is a breach of the principle of equal pay for men and women, the comparison must cover a relatively significant number of individuals so as to exclude purely fortuitous phenomena.