In this paper I review some of the lessons to be learned from long term soil experiments, and existing and new networks of long term experiments, before reviewing some of the novel applications to which they have been put, and some of the drawbacks inherent in long term experiments.

Long term soil experiments have provided many answers that it would not have been possible to obtain from other sources, including vital information on the impacts of land-use change, climate change, pollution, on parameters ranging from soil carbon to net primary production to biodiversity.

They have also been used to evaluate and test models and to test sensitivity and uncertainty of model outputs. Because many experiments were set up many years ago, they have drawbacks, often including lack of replication and few independent samples with to assess statistical power / uncertainty. These potential short-comings are reviewed.

Nevertheless, long term soil experiments potentially now have an important new role in forming benchmark sites as part of integrated national carbon accounting systems; examples are given from Canada. To fulfil this new role, many experiments will need to adapt, but the challenge is a worthy one, giving a new long term future to experiments in a world with increasingly short term priorities.