BREYER, Circuit Judge.

Appellants, representing the Commonwealth of Massachusetts, seek to overturn a district court order requiring them to develop, and to pay for, a program of legal assistance for mentally ill and retarded people. The district court, in connection with its ongoing supervision of a consent decree, ordered the Commonwealth to create an independent legal advocacy program that would represent mentally ill or retarded clients on issues arising out of a client's "disability," "prior . . . institutionalization," or "current institutionalization." After examining the language of the decree, its background, its purposes, and other relevant law, we agree with appellants that the court has no legal authority to require the Commonwealth over its objection to pay for a broad legal services program. We therefore vacate the district court's order.

I

The consent decree in this case represents the negotiated settlement of a suit filed in 1976 as a class action in federal district court by nine residents of Northampton State Hospital in Western Massachusetts. The plaintiffs, appellees here, claimed that the Commonwealth's programs for treating mentally ill and retarded people violated state and federal law (constitutional and statutory) which, in their view, required treatment in less restrictive, more normal surroundings. After two years of negotiation, the parties agreed to enter a consent decree. The Commonwealth — represented by the Governor, the Attorney General, the Secretary of Administration and Finance, the Secretary of Human Services, and the Commissioner of the Department of Mental Health — did not admit any violation of law. But in the decree it promises to provide a "less restrictive" system of mental health care. It undertakes to provide community mental health services, offering less restrictive residential facilities and nonresidential treatment programs, with the aim of allowing hospitalized patients to leave Northampton State Hospital and find a place to live, and to obtain treatment, in their communities. The decree is lengthy and detailed; it explicitly leaves some issues undecided and open to future resolution. It provides for ongoing court supervision with the assistance of a court-appointed monitor. It has been the subject of previous litigation in this court, see, e.g., Brewster v. Dukakis, 675 F.2d 1 (1st Cir. 1982).

This particular appeal arises out of the decree's arrangements for resolving one of the controversies explicitly left undecided, namely, the appropriate type, role, and financing of a system of legal advocacy for the mentally ill and retarded. Although the parties could not agree about these matters in 1978, they wrote into the decree a provision that dealt with the issue as follows:

¶ 59. The Monitor will investigate and determine the necessity for trained, independent advocates to assist clients in the protection of their rights as set forth in relevant statutes, regulations, and the provisions of this Decree, including the attachments hereto. The Monitor will submit its recommendation to the parties and the Court by January 1, 1980, on the appropriate role of and funding for such independent advocacy. The plaintiffs and defendants agree to cooperate in investigating sources of funding and in seeking federal funds to establish and maintain an advocacy system, to the extent such is determined to be appropriate.

Pursuant to this paragraph, the Monitor asked an expert consultant, a lawyer with special qualifications and experience in this area, to undertake a study of the problem. The consultant prepared a lengthy report, and submitted it to the Monitor in September 1980. This report appears to constitute the basis for the order under challenge.

The report begins by setting out many rights that the author believes are conferred upon mentally ill people by law, including the decree, administrative regulations, statutes and constitutions. Some of these rights are highly general, such as rights to treatment, to a nonrestrictive environment, to decent facilities, to fair procedural rules. Others involve nondiscrimination, such as the right of equal access to educational, recreational, transportation, and medical services. Still others are detailed and specific, such as rights to stationery, postage, storage space, and access to a telephone. As these examples suggest, the list of rights is extensive, ranging from practical issues, such as a decent diet, to individual rights, such as treatment with dignity. The report goes on to make a powerful case that the 13 pages of specific rights which it lists as those of the mentally ill may remain only paper promises, unless the mentally ill are also provided with independent advocates who can represent them individually. The report's 200 pages argue clearly and strongly for a general, state-financed, system of independent advocates to secure these rights for the mentally ill and retarded.

In January 1981, the Monitor, acting pursuant to ¶ 59, made recommendations more modest, but roughly similar to those that the consultant suggested. He noted that independent legal advocates are needed and that, after June 1982, it will be difficult for many existing advocacy organizations to find funding. He concluded that the state should pay for at least a few lawyers to act as the nucleus of a general, independent legal advocacy system.

Subsequently, the Commonwealth asked the court to declare that it had no legal obligation to pay for such a program. The court held an evidentiary hearing. The consultant and others testified. Appellees submitted affidavits, for example, from patients whom legal advocates had helped to obtain such benefits as better treatment and vocational training. Appellants submitted evidence that their existing program of "service coordinators" meets those needs for which the Monitor recommended independent legal advocates. But, the Commonwealth's main argument was a legal one — that it could not be required to pay for the legal advocacy program, because the law imposes no such obligation.

The district court, on August 19, 1981, rejected the Commonwealth's argument and accepted the Monitor's recommendations. It ordered the Commonwealth to present a specific detailed plan for carrying out those recommendations. Appellants appeal from that order.

II

Although appellants argued in the district court that the Commonwealth's "service coordinator" program was adequate to secure the rights of the mentally ill, they will concede for purposes of this appeal that an independent legal advocacy program is highly desirable. They do not take issue with the soundness of the recommendations of the expert consultant or the Monitor. They argue only that, as a matter of law, the district court lacks the power to force the Commonwealth to pay for the recommended program. And that question — not the desirability of the program — is the sole issue that we consider.

The district court found the source of its power in the decree itself. We have examined the decree and the record in detail, keeping in mind that the district court is more familiar with the background of the litigation than are we. Cf. Brown v. Neeb, 644 F.2d 551, 558 n. 12 (6th Cir. 1981). Nonetheless, we have been unable to find adequate authority in the decree for the court's order. Nor can we find adequate authority elsewhere in the law. If the decree provides this authority, it must be found in some combination of language, structure or purpose. A discussion of each of these aspects of the decree, in turn, should indicate why we are unable to find the requisite authority.

1. Language. The language of the decree, if given its ordinary meaning, does not authorize the court to order the Commonwealth to fund a general legal advocacy program for the mentally ill. Paragraph 59, which deals specifically with legal advocacy, instructs the Monitor to " investigate and determine" the need for "trained, independent advocates." He is to "submit" his " recommendation to the parties and the Court." (Emphasis added.) The paragraph goes on to state that the parties will "cooperate in investigating sources of funding and in seeking federal funds." But it says no more. It does not specifically authorize the court to order the parties to accept the Monitor's recommendations. Cf. New York State Association for Retarded Children, Inc. v. Carey, 596 F.2d 27, 33, 37 (2d Cir.), cert. denied, 444 U.S. 836, 100 S.Ct. 70, 62 L.Ed.2d 46 (1979).

The court relied for its power upon ¶ 59 in combination with ¶¶ 5 and 6 of the decree. Yet, the language of these paragraphs, if given its ordinary meaning, also seems inadequate. The relevant sentences of ¶ 5 (which appears in the "general" section of the decree) read as follows:

This Decree includes a final, substantive resolution of most of the issues raised by this litigation. To the extent that current information renders final, substantive settlement of an issue impracticable, the parties have agreed upon a schedule and process for collaboration and/or negotiation, supervised by the Court, which the parties believe can lead to a final resolution of the issue in a fair and reasonable manner.

This language suggests that the parties intended to resolve through negotiation or collaboration certain issues that the decree left open. This expectation seems reasonable given the fact (as the briefs point out) that relations between the parties were good: both sides believed in, and were working to achieve, "deinstitutionalization" (the main object of the decree); and negotiation has led to resolution of most of the other undecided issues. Moreover, this language says only that the court shall "supervise" the "process for collaboration and/or negotiation." It does not say that the court shall "resolve" the issues subject to that process. Courts, as well as administrative bodies, are familiar with the difference between encouraging parties to settle a dispute (or supervising a settlement process) and deciding a dispute over objection. There is not here any evidence of "bad faith" by a party that could transform the one power into the other.

Paragraph 6 (also contained in the "general" section of the decree) reads as follows:

Jurisdiction is retained by the Court until further order, to enable any party to apply at any time for such further orders as may be necessary or appropriate for the interpretation, implementation, enforcement, or modification of the terms of this Decree and for supervision and approval of the resolution of issues left for further planning and negotiation.

The first part of this sentence does not grant the needed authority. Even if one viewed this order as one "appropriate for the . . . implementation . . . of the terms" of the decree, one would still have to find elsewhere in the decree specific "terms" that the order implements. Nor does the order purport to be one of "modification" of terms. The last clause of the sentence gives the court the power of "supervision and approval of the resolution of issues left for further planning and negotiation . . . ." (Emphasis added.) It does not speak of the court's powers of "supervision and approval and resolution" of issues left for future settlement. Thus, it is most naturally read as specifying the court's authority to incorporate into the decree, and to implement, the results of negotiations on open issues, not as granting the court authority to resolve those issues itself.

Paragraph 50, to which appellees also point, states that if unforeseen difficulties arise which "significantly threaten implementation, the problem may be brought to the Court for its attention." There is no indication that we are dealing here with an "unforeseen difficulty." But, even if this paragraph were relevant, its language does not confer the power at issue.

In sum, the language of the decree's "general" provisions, while recognizing the existence of major open issues, cannot, without straining, be read automatically to grant the court a general power to resolve all open issues, other than in conformity with procedures specified elsewhere. And, the more specific provision relevant here ( ¶ 59) allows the Monitor to investigate and to recommend, but it does not state that the Monitor or the court shall resolve the matter of a broad, general, independent, legal services program for the mentally ill.

Nonetheless, these clauses do not forbid court resolution. And, there is, at least arguably, some ambiguity in some of the language. We therefore go on to see whether there are special reasons, arising out of the decree's structure or purposes, or out of the parties' likely intent, for allowing the district court, nonetheless, to read the needed authority into this language.

2. Structure. The "structural" argument rests on an analogy to three other sets of provisions that concern issues deliberately left open. These other provisions arguably bear upon the proper interpretation of the language just discussed. The first set of these provisions concerns the powers of the Monitor. Paragraph 55, for example, states that "an individual complaint or problem" of a client or resident may be brought to the attention of the Monitor "for appropriate action." He may make "a recommendation to the defendants" to resolve the problem. His "action on individual complaints or problems will be considered final." Paragraph 56 gives the Monitor "authority to make recommendations with regard to implementation of the Decree if . . . the Monitor believes that the defendants are not in compliance with the Decree [and certain other conditions are fulfilled] . . . ." The "recommendations" are binding on the parties unless they file an objection with the court and request a hearing. Paragraph 57 allows the monitor to "make recommendations on unresolved issues and propose modifications or revisions in the Decree to facilitate compliance with its fundamental purposes." Paragraph 58 allows a party to ask the Monitor "for a specific action or recommendation," and a party objecting to the "Monitor's response may request a hearing before the Court."

These provisions do not shed much light on the present problem. Paragraphs 55 and 56 provide for binding recommendations. But they deal with specific circumstances (such as "individual" grievances) which the parties apparently agree are not present here. And, since they use specific language to make a recommendation binding where they do apply, they are of little help in deciding how to treat a "recommendation" where they do not apply. The last two paragraphs — ¶¶ 57 and 58 — also speak of a Monitor's "recommendation" but they do not suggest that a "recommendation" is binding or that the court has power to make it so.

Second, appellees point to several provisions that both deal with issues left open and also seem to foresee the court's having the power, if necessary, to resolve them. Paragraphs 32 and 33 state that appellant Department "will draft and promulgate regulations, . . . governing residential alternatives, nonresidential programs, Individual Service Plans, and due process procedures for residents moving from the Hospital to a community setting." The regulations "will be promulgated," they "will . . . be incorporated by reference" in the decree, and they "will be jointly drafted through a negotiation process between the plaintiffs and the Department." Similarly, ¶ 35 provides that "the parties" (presumably acting together) "will prepare . . . a plan for orientation, retraining, and ongoing development of Hospital staff and employees of community programs . . . [which] plan will be incorporated by reference in this Decree after review and approval by all parties." Paragraph 43 provides that "[t]he parties . . . will produce, through a negotiation process, a reorganization and phase-down plan" for the Northampton State Hospital, which plan "will be included by reference in this Decree." Appellees argue that these provisions — dealing with regulations, personnel training, and hospital reorganization — do not expressly grant the court authority to resolve the issue; yet, it is reasonable to imply such authority despite silence. Thus, one might imply an intent to grant the court authority to resolve other issues — indeed all issues left open for future resolution.

This argument, however, is weak. If these provisions bestow upon the court the power to impose solutions, they do so in virtue of 1) their rather specific language (the regulations, the training plan, the phase-down plan, all "will be" written and incorporated), and 2) the need to resolve these details of "deinstitutionalization" (regulations, personnel training, reorganization of the hospital) to achieve the decree's agreed upon, major, "deinstitutionalization" objective. Yet, it is as to these two key points that the relation of these paragraphs to the legal services paragraph ( ¶ 59) breaks down. An inspection of the language of ¶ 59 reveals that it is far less specific in foreseeing a resolution of the issue than are ¶¶ 32, 33, 35 and 43. And, as will be elaborated below, the legal services program at issue in ¶ 59 is not a "deinstitutionalization" detail, logically implied by a commitment to the objective. Rather, it constitutes a major, separable program which could in good faith be opposed (or favored) by one committed to achieving "deinstitutionalization."

Third, appellees point to ¶ 16, which provides the closest analogy to ¶ 59. It concerns the problem of what to do with the "remaining" population at Northampton State Hospital — those who appear to need "a secure treatment setting at an appropriate location." It states that, "[i]f the parties cannot agree on a resolution of" this issue, "they will report to the Court." This paragraph is very much like ¶ 59, because it does not say what the court can or cannot do. The matter is simply to be referred to the court. Moreover, like a general legal services program, resolution of issues related to those who cannot be "deinstitutionalized" is not obviously essential to the success of deinstitutionalization. The problem with pointing to ¶ 16, however, is that ¶ 16 raises exactly the same question at issue here. It is no more clear than ¶ 59 that the court is to have the power to resolve differences if the parties fail to agree. Nothing in the record before us shows that it gives this power to the court; thus, it is an analogy that shows little.

In sum, the "structural" argument by analogy to other provisions in the decree fails to show the necessary delegation of power. Those provisions that appear to grant the power at issue in respect of their own particular subject matter ( ¶¶ 55, 56, 32, 33, 35, 36, 43) are not similar enough to ¶ 59 to be convincing. The provision that seems most similar ( ¶ 16) does not clearly delegate the necessary power.

3. Purpose. The key question is whether the main purpose of the decree — deinstitutionalization — requires a general legal services program for the mentally ill or at least requires that the court have the power to create such a program. If so, one might argue that the parties intended to give the court this power on the principle that he who wills the end must will the means. After examining the record, however, we do not believe the court power here at issue can be implied from the basic purposes of the consent decree.

As described in the decree itself, in letters written by appellees' counsel, and in court opinions, the decree's basic purpose is "deinstitutionalization" — that is to say, removing patients from Northampton State Hospital and providing for their care in less restrictive, more normal, surroundings. The aim of the system described in the court's order, in the Monitor's recommendation, and in the consultant's report, however, is different. It is to provide legal representation on legal issues arising generally out of the "clients' disability" as well as their "prior or current institutionalization." The opinion, recommendations, and report make clear that the relevant range of legal issues on which representation is sought is vast, encompassing everything from job discrimination to surgical treatment. Clearly, legal representation and "deinstitutionalization" are related; the former can help achieve the latter. But, the relationship between a general legal advocacy program and deinstitutionalization is neither close enough nor special enough to conclude that parties intending the latter must have intended the former, or must have intended to allow the court to order it in the event of disagreement.

We reach this conclusion primarily for the reason just stated. The legal services program described summarily in the court's order extends well beyond problems arising out of "deinstitutionalization;" it is not a program tailored to decree enforcement; it is a program for general representation of the mentally ill; it thus could not constitute a necessary part of the decree's basic deinstitutionalization objective. Moreover, it does not seem inevitable to us to infer an intent to resolve definitively the "legal services" issue (unlike the "regulations," "training" or "hospital reorganization" issues) from the basic deinstitutionalization commitment. At least one of the parties, the Commonwealth, has never believed (given its own "service coordinators") that independent legal advocacy is an absolute necessity. It still argues that, even if independent legal advocacy is desirable, as a practical matter funds are limited, and whatever money is available is better spent on treatment, or other patient needs. Cf. Brewster v. Dukakis, supra. Whether or not its arguments are correct, the fact that they are made, and that they are reasonable, suggests that the Commonwealth did not intend, and that it need not have intended, that the legal services issue be resolved, as part of its commitment to deinstitutionalization. Thus, even if we interpret the language of the decree flexibly "in light of the need to achieve . . . [its] basic purposes," Massachusetts Association for Retarded Citizens, Inc. v. King, 668 F.2d 602, 607-08 (1st Cir. 1981), those purposes do not imply a grant by the parties to the court of the power to order the Commonwealth, over its objection, to pay for a general legal services program in Western Massachusetts to represent mentally ill and retarded persons on all issues related to their disabilities.

We specifically confine our holding to the general legal services program referred to in the district court's order. Some of the statements made at oral argument suggest that the order's program might also cover the legal cost of representing plaintiffs in this and other proceedings directly related to the implementation and enforcement of the decree. But we are not considering an order that deals specifically with these costs. The order before us is broader. An order narrowly aimed at legal representation in this case itself would raise quite different issues from those argued here, in terms of both expectations and needs. And, by vacating this order, we do not intend to prejudge the determination of the lawfulness of a more narrow order aimed at legal representation in this case itself.

4. Other sources of law. Finally, we do not see how the court's powers can be augmented by references to "general equitable," or to other, powers, to cure violations of the Constitution, for the simple reason that there has been no finding of any such violation. The appellees' legal claims were not adjudicated; the Commonwealth did not concede they were valid; nor has it been persuasively shown that the requisite findings can be inferred from the circumstances. Cf. NAACP v. Beecher, 679 F.2d 965 (1st Cir. 1982); Brown v. Neeb, 644 F.2d at 561-63. See Youngberg v. Romeo, ___ U.S. ___, ___ n. 25, 102 S.Ct. 2452, 2460 n. 25, 73 L.Ed.2d 28 (1982) ("A federal court, of course, must identify a constitutional predicate for the imposition of any affirmative duty on a state."). Thus, we cannot here accept as controlling authority, cases such as Halderman v. Pennhurst State School and Hospital, 446 F. Supp. 1295, 1325, 1327 (E.D.Pa. 1977), aff'd on federal statutory grounds, 612 F.2d 84, 113 (3d Cir. 1979), rev'd, 451 U.S. 1, 101 S.Ct. 1531, 67 L.Ed.2d 694 (1981), where violations of law were found.

For these reasons, we vacate the order of the district court and remand for proceedings consistent with this decision.

Vacated and remanded.