The first national deal for electronic data is now six years old and was for the citation indices of the Institute for Scientific Information. This first and still flagship deal continues to see ever increasing growth in usage. This was followed by a range of essentially bibliographic products ranging from Beilstein’s Crossfire service to the International Bibliography of the Social Sciences. These deals were negotiated by CHEST and allowed the Higher Education community to use its purchasing power collectively to persuade publishers that co-operative subscription based contracts were a better model than transaction based contracts. Although all universities contribute towards the costs of each dataset to which they subscribe, this has allowed us both to reach a larger set of universities than would have installed the services individually and, perhaps more importantly, to attempt the provision of standard interfaces.

The data is provided from five national data centres selected competitively from amongst the universities. There are several reasons for this, which collectively form a national policy. Firstly, distribution of data over a limited number of centres allows some concentration of investment and some management of traffic flows on the network. Secondly, this allows us to concentrate human resources and skills in areas as varied as the management of very large datasets and the design of interfaces. Thirdly, we have found that publishers prefer dealing with what they see as a group of professional dataset suppliers. Fourthly, we are reluctant to rely entirely on North America as a data source and wish to see a strong European information industry. We therefore see national value in having a trained cadre of workers with relevant skills.

As well as purchasing data we have recently set up a national digitisation centre at the University of Hertfordshire. In part this subsidises data creation and in part it will explore the economics of digitisation and look at relevant standards issues. It is expected to concentrate on the digitisation of out of copyright material or on partnership deals with publishers, rather like the JSTOR project in the United States. This emphasises part of the philosophy, which wishes to see the creation of a critical mass of data in academic subjects, thus ensuring that digitised collections become an unremarkable part of everyday teaching and research rather than being exceptional "toys".

We have been concerned that all disciplines should benefit from electronic libraries and so a significant programme has been created to support special library collections in the humanities. Although much of this has gone for the cataloguing of older and archive material, a significant amount has gone to digital preservation of everything from mediaeval manuscripts to documentary film. It is an express condition of grants that the resources must be made publicly available.

The converse of this is that substantial quantities of "raw" satellite data are purchased and stored at one of the data centres, while very large government datasets such as the Census are held at another. This both provides a rich resource for the community at large and again concentrates expertise in specialised centres.

There is a common and justified perception that there is too much useless, irrelevant and misleading information on the Internet. Several resource discovery projects have been set up covering faculty level areas such as engineering and medicine. These will identify and catalogue relevant internet resources as well as ensuring their quality and availability. Although individual universities can (and no doubt will) develop their own resource discovery services, these are sufficiently ill understood that it has been felt that a national initiative will best explore and develop standards and allow the UK to take part in international developments. It has been felt that a central push will provide larger benefits than fragmented individual efforts.

Perhaps the most important element of the programme lies in training for cultural change. Large sums have been allocated for the development of training programmes which will develop the professional skills of library and information workers.

The most recent material provided is a large range of electronic journals from three publishers – Academic Press, Blackwell Scientific and the Institute of Physics Press. Although each deal is a little different, not least to allow different models to be tested, they are intended to provide the complete list of each publisher in electronic form. Discussions were difficult and protracted and the publishers much less able than they had thought to provide the journals electronically. Individual universities have also learned valuable lessons about the cost of supporting electronic journals as a mass service. These range from training needs to the provision of the correct browsers and software on every computer in the university. The general lesson learned is that purchase costs is the least important issue and support the most important.

This large explanation is intended to convey several points of which the major one is the coherence and extent of the programme of content provision. Much effort is expended in looking at how to deal with and make deals with large scientific publishers. We have felt this to be important but misguided. Companies such as Elsevier or Springer are the exception. Small learned societies are the norm for scholarly communication, while much non-commercial and out of copyright material is of value in research. It has seemed important to ensure a rich and varied set of content rather than simply strike a set of commercial deals. In addition this has allowed sensible exploration of infrastructure issues from interface design to network traffic management.

We have also been clear that information should be available in all disciplines and at all levels. We are committed to creating information literate graduates and they require access to the resources they will use when they move into employment. A skilled workforce is crucial to the future of the country. Likewise we believe that the digital library is important in all disciplines and we must guard against allowing the STM publishing model to drive all debate. Theologians are as important to the universities as chemists.

We have been eager to create a critical mass of content, discipline by discipline. Much digital library experimentation has been to prove concepts. There is general agreement that the real test then comes in scaling up experiments to operational services and in making those services essential scholarly tools rather than experimental toys.

Finally we have felt it important to demonstrate value for money. A large evaluation programme has been commissioned. Ultimately we are expending taxpayers money and we wish this not only to be done, but to be seen to be done, effectively, economically and efficiently.

The critical area is undoubtedly support. Publishers have worked in a climate where support was barely needed beyond the odd replacement copy of a book. But the complex nexus of data, telecommunications and software requires extensive support which may be summed up in the maxim, "User-friendly systems are not".

The other critical lesson is that of partnership. Many libraries have tried to build their own digital equivalent. We believe that there are lessons to be learnt and shared together rather than alone. Similarly co-operating libraries cannot alone afford to create and/or provide all the digital content which will be required in future. We need partnerships with publishers and learned societies. We have lessons, technical skills and experience to offer them as they move into the cold harsh economics of the digital world.