To analyze science as practice and culture has become, since the early 1970s, the object of the new history and sociology of science. Hence, historians and sociologists pay now more attention to the role of experiment in science. In order to study experiments we need to think more carefully about instruments, apparatus and their use. In this article I put forward a method which allows to do both, to study the materiality of experiment as well as the activities involved in (...) the production of experimental results: The replication of an experiment, ie. the reworking of historical experiments with a replica as close to the original as possible. A study of the experiments jointly done by James Joule, and William Thomson in 1852 will demonstrate what might be learned by this method about an experiment. The origins of instruments and the apparatus used in these experiments, their historical context and use will be discussed in detail. With the help of this case study I will try to show that the replication of experiments has much to contribute to an enhanced understanding of experimental practices. (shrink)