Sir

Recent correspondents (see for example, K. Alverson and M. Eakin, and M. Boulter, Nature 412, 269; 2001) emphasize the importance of general repositories for scientific data. They point out the lack of appropriate financing for scientific data management and the need for thematically related scientific-information systems to be networked, which is technically feasible.

The current problem resides deeper. To enact global agreements such as the Kyoto protocol, for example, decision-makers are required to pursue policies that cover many different, yet plausible, estimates of the likelihood of alternative, future climate development. These climate-change scenarios are frequently derived from climate models, whose correctness can be measured only by access to raw data.

The necessary database infrastructure was created as early as the 1950s, when the International Council of Scientific Unions set up the World Data Center (WDC) system (http://www.ngdc.noaa.gov/wdc). These centres are for the international exchange of solar, geophysical and related environmental data on a long-term basis, and are to assist principal investigators (PIs) in broad data management. Yet there are no international regulations requiring scientists to store results as raw data and accompanying meta-information in this or any other publicly accessible archive.

Some scientific journals and funding agencies encourage PIs or authors to submit raw data or support data sharing. Yet the US National Science Foundation (NSF) is the only funding organization that insists on this. In its information on submitting climate-related proposals for its Earth System History Program (http://www.nsf.gov/pubs/2000/nsf0011/nsf0011.html), the NSF says: “Each proposal must adhere to the USGCRP [US Global Change Research Program] data management policy ... and the policies applying to recipients of Federal funding in the geosciences. Unless otherwise specified in the proposal, the [PI] will be responsible for ensuring that all data generated by the funded project will be documented and submitted to the World Data Center”.

Our example of climate-change models is but one of many where raw data are needed for accurate scientific assessment in the long term. An internationally binding regulation that adheres to WDC principles is required. It should guarantee that all scientific data are archived and freely available. Because some PIs still refuse to archive data individually in appropriate databases such as the WDC, still refuse to make their published data publicly available as raw data, and still ignore the potential benefit from networking through WDCs, additional financial and temporal effort must be spent to gain access to these data.

Until a free-access global repository can be set up, this may be the best first step. The long-term goal must be to secure the world's stock of valuable data, and to overcome practical difficulties in so doing.