Re S (Minors) (Care order: implementation of care plan)  UKHL 10:  2 All ER 192
LORD NICHOLLS OF BIRKENHEAD: My Lords,
 These appeals concern the impact of the Human Rights Act 1998 on Parts III and IV of the Children Act 1989. The Court of Appeal (see Re W and B; Re W (Care Plan)  EWCA Civ 757,  2 FLR 582) (Thorpe, Sedley and Hale LJJ) made, in the words of Thorpe LJ, two major adjustments and innovations in the construction and application of the Children Act 1989. The principal issue before your Lordships' House concerns the soundness of this judicial initiative....
 Two preliminary points can be made at the outset. First, a cardinal principle of the Children Act 1989 is that when the court makes a care order it becomes the duty of the local authority designated by the order to receive the child into its care while the order remains in force. So long as the care order is in force the authority has parental responsibility for the child. The authority also has power to decide the extent to which a parent of the child may meet his responsibility for him: s 33. An authority might, for instance, not permit parents to change the school of a child living at home. While a care order is in force the court's powers, under its inherent jurisdiction, are expressly excluded: s 100(2)(c) and (d). Further, the court may not make a contact order, a prohibited steps order or a specific issue order: s 9(1).
 There are limited exceptions to this principle of non-intervention by the court in the authority's discharge of its parental responsibility for a child in its care under a care order. The court retains jurisdiction to decide disputes about contact with children in care: s 34 of the Children Act 1989. The court may discharge a care order, either on an application made for the purpose under s 39 or as a consequence of making a residence order (ss 9(1) and 91(1)). The High Court's judicial review jurisdiction also remains available.
 These exceptions do not detract significantly from the basic principle. The Children Act 1989 delineated the boundary of responsibility with complete clarity. Where a care order is made the responsibility for the child's care is with the authority rather than the court. The court retains no supervisory role, monitoring the authority's discharge of its responsibilities. That was the intention of Parliament.
 Consistently with this, in Kent County Council v C  Fam 57, 993] 1 FLR 308 Ewbank J decided that the court has no power to add to a care order a direction to the authority that the child's guardian ad litem should be allowed to have a continuing involvement, with a view to his applying to the court in due course if thought appropriate. In Re T (A Minor) (Care Order: Conditions)  2 FLR 423 the Court of Appeal rightly approved this decision and held that the court has no power to impose conditions in a care order. There the condition sought by the child's guardian was that the child should reside at home.
 This cardinal principle of the Children Act 1989 represented a change in the law. Before the Children Act 1989 came into operation the court, in exercise of its wardship jurisdiction, retained power in limited circumstances to give directions to a local authority regarding children in its care. The limits of this jurisdiction were considered by your Lordships' House in A v Liverpool City Council  AC 363, (1981) 2 FLR 222 and Re W (A Minor) (Wardship: Jurisdiction)  AC 791, sub nom Re W (A Minor) (Care Proceedings: Wardship)  FLR 879. The change brought about by the Children Act 1989 gave effect to a policy decision on the appropriate division of responsibilities between the courts and local authorities. This was one of the matters widely discussed at the time. A report made to ministers by an inter-departmental working party Review of Child Care law (September 1985) drew attention to some of the policy considerations. The particular strength of the courts lies in the resolution of disputes: its ability to hear all sides of a case, to decide issues of fact and law, and to make a firm decision on a particular issue at a particular time. But a court cannot have day-to-day responsibility for a child. The court cannot deliver the services which may best serve a child's needs. Unlike a local authority, a court does not have close, personal and continuing knowledge of the child. The court cannot respond with immediacy and informality to practical problems and changed circumstances as they arise. Supervision by the court would encourage 'drift' in decision making, a perennial problem in children cases. Nor does a court have the task of managing the financial and human resources available to a local authority for dealing with all children in need in its area. The authority must manage these resources in the best interests of all the children for whom it is responsible.
 The Children Act 1989, embodying what I have described as a cardinal principle, represents the assessment made by Parliament of the division of responsibility which would best promote the interests of children within the overall care system. The court operates as the gateway into care, and makes the necessary care order when the threshold conditions are satisfied and the court considers a care order would be in the best interests of the child. That is the responsibility of the court. Thereafter the court has no continuing role in relation to the care order. Then it is the responsibility of the local authority to decide how the child should be cared for.
 My second preliminary point is this. The Children Act 1989 has now been in operation for 10 years. Over the last 6 years there has been a steady increase in the number of children looked after by local authorities in England and Wales. At present there are 36,400 children accommodated under care orders, compared with 28,500 in 1995, an increase of 27%. In addition local authorities provide accommodation for nearly 20,000 children under s 20 orders (children in need of accommodation). A decade's experience in the operation of the Children Act 1989, at a time of increasing demands on local authorities, has shown that there are occasions when, with the best will in the world, local authorities' discharge of their parental responsibilities has not been satisfactory. The system does not always work well. Shortages of money, of suitable trained staff and of suitable foster carers and prospective adopters for difficult children are among the reasons. There have been delays in placing children in accordance with their care plans, unsatisfactory breakdown rates and delays in finding substitute placements.
 But the problems are more deep-seated than shortage of resources. In November 1997 the Government published Sir William Utting's review of safeguards for children living away from home. Mr Frank Dobson, then Secretary of State for Health, summarised his reaction to the report:
'It covers the lives of children whose home circumstances were so bad that those in authority, to use the jargon, took them into care. The report reveals that in far too many cases not enough care was taken. Elementary safeguards were not in place or not enforced. Many children were harmed rather than helped. The review reveals that these failings were not just the fault of individuals -- though individuals were at fault. It reveals the failure of a whole system.'
 In autumn 1998 the Government published its response to the children's safeguards review (Cm 4105) and launched its Quality Protects Programme, aimed at improving the public care system for children. Conferences have also been held, and many research studies undertaken, both private and public, on particular aspects of the problems. Some of the problems were discussed at the bi-annual President's Inter-disciplinary Conference on Family Law 1997, attended by judges, child psychiatrists, social workers, social services personnel and other experts. The proceedings of the conference were subsequently published in book form, Divided Duties: Care Planning for Children within the Family Justice System (Family Law, 1st edn, 1998). The sharpness of the divide between the court's powers before and after the making of a care order attracted criticism. The matters discussed included the need for a care plan to be open to review by the court in exceptional cases. One suggestion was that a court review could be triggered by failure to implement 'starred' key factors in the care plan within specified time-scales. The guardian ad litem would be the appropriate person to intervene.
 This was the source of the innovation which found expression in the judgments of the Court of Appeal in the present appeals. The House was informed by counsel that the starred milestones guidance given by the Court of Appeal was not canvassed in argument before the court. This guidance appeared for the first time in the judgments of the court.
 The jurisprudential route by which the Court of Appeal found itself able to bring about this development was primarily by recourse to s 3 of the Human Rights Act 1998. Hale LJ said in Re W and B; Re W (Care Plan)  EWCA Civ 757,  2 FLR 582, paras -:
'Where elements of the care plan are so fundamental that there is a real risk of a breach of Convention rights if they are not fulfilled, and where there is some reason to fear that they may not be fulfilled, it must be justifiable to read into the Children Act 1989 a power in the court to require a report on progress . . . the court would require a report, either to the court or to the guardian ad litem (in future to CAFCASS) who could then decide whether it was appropriate to return the case to court . . .
(When) making a care order, the court is being asked to interfere in family life. If it perceives that the consequence of doing so will be to put at risk the Convention rights of either the parents or the child, the court should be able to impose this very limited requirement as a condition of its own interference.' (my emphasis)
Section 3 of the Human Rights Act
 The judgments in the Court of Appeal are a clear and forceful statement of the continuing existence of serious problems in this field. In the nature of things, courts are likely to see more of the cases which go wrong. But the view, widespread among family judges, is that all too often local authorities' discharge of their parental responsibilities falls short of an acceptable standard. A disturbing instance can be found in the recent case of Re F; F v Lambeth London Borough Council  1 FLR 217. Munby J said, at para  of his judgment, that the 'blunt truth is that in this case the State has failed these parents and these boys'.
 It is entirely understandable that the Court of Appeal should seek some means to alleviate these problems: some means by which the courts may assist children where care orders have been made but subsequently, for whatever reason, care plans have not been implemented as envisaged and, as a result, the welfare of the children is being prejudiced. This is entirely understandable. The courts, notably through their wardship jurisdiction, have long discharged an invaluable role in safeguarding the interests of children. But the question before the House is much more confined. The question is whether the courts have power to introduce into the working of the Children Act 1989 a range of rights and liabilities not sanctioned by Parliament.
 On this I have to say at once, respectfully but emphatically, that I part company with the Court of Appeal. I am unable to agree that the court's introduction of a 'starring system' can be justified as a legitimate exercise in interpretation of the Children Act 1989 in accordance with s 3 of the Human Rights Act 1998. Even if the Children Act 1989 is inconsistent with Art 6 or 8 of the Convention, which is a question I will consider later, s 3 does not in this case have the effect suggested by the Court of Appeal.
 Section 3(1) of the Human Rights Act 1998 provides:
'So far as it is possible to do so, primary legislation . . . must be read and given effect in a way which is compatible with the Convention rights.'
This is a powerful tool whose use is obligatory. It is not an optional canon of construction. Nor is its use dependent on the existence of ambiguity. Further, the section applies retrospectively. So far as it is possible to do so, primary legislation 'must be read and given effect' to in a way which is compatible with Convention rights. This is forthright, uncompromising language.
 But the reach of this tool is not unlimited. Section 3 of the Human Rights Act 1998 is concerned with interpretation. This is apparent from the opening words of s 3(1): 'so far as it is possible to do so'. The side heading of the section is 'Interpretation of legislation'. Section 4 (power to make a declaration of incompatibility) and, indeed, s 3(2)(b) presuppose that not all provisions in primary legislation can be rendered Convention compliant by the application of s 3(1). The existence of this limit on the scope of s 3(1) has already been the subject of judicial confirmation, more than once: see, for instance, Lord Woolf CJ in Poplar Housing and Regeneration Community Association Ltd v Donoghue  EWCA Civ 595,  QB 48, sub nom Donoghue v Poplar Housing and Regeneration Community Association Ltd  2 FLR 284, para  and Lord Hope of Craighead in R v Lambert  UKHL 37,  3 WLR 206, paras -.
 In applying s 3 courts must be ever mindful of this outer limit. The Human Rights Act 1998 reserves the amendment of primary legislation to Parliament. By this means the Act seeks to preserve parliamentary sovereignty. The Act maintains the constitutional boundary. Interpretation of statutes is a matter for the courts; the enactment of statutes, and the amendment of statutes, are matters for Parliament.
 Up to this point there is no difficulty. The area of real difficulty lies in identifying the limits of interpretation in a particular case. This is not a novel problem. If anything, the problem is more acute today than in past times. Nowadays courts are more 'liberal' in the interpretation of all manner of documents. The greater the latitude with which courts construe documents, the less readily defined is the boundary. What one person regards as sensible, if robust, interpretation, another regards as impermissibly creative. For present purposes it is sufficient to say that a meaning which departs substantially from a fundamental feature of an Act of Parliament is likely to have crossed the boundary between interpretation and amendment. This is especially so where the departure has important practical repercussions which the court is not equipped to evaluate. In such a case the overall contextual setting may leave no scope for rendering the statutory provision Convention compliant by legitimate use of the process of interpretation. The boundary line may be crossed even though a limitation on Convention rights is not stated in express terms. Lord Steyn's observations in R v A (No 2)  UKHL 25,  1 AC 45, para  are not to be read as meaning that a clear limitation on Convention rights in terms is the only circumstance in which an interpretation incompatible with Convention rights may arise.
 I should add a further general observation in the light of what happened in the present case. Section 3 directs courts on how legislation shall, as far as possible, be interpreted. When a court, called upon to construe legislation, ascribes a meaning and effect to the legislation pursuant to its obligation under s 3, it is important the court should identify clearly the particular statutory provision or provisions whose interpretation leads to that result. Apart from all else, this should assist in ensuring the court does not inadvertently stray outside its interpretation jurisdiction....