What can we learn from repeating historical experiments?
In recent decades a good number of historians of science have been engaged in the replication of experiments from the history of science. With the work presented in this paper I am joining this existing trend, but my work has quite a different set of aims from most historians'. To put it simply: my main purpose is to learn about nature, not about past scientists or the precise conditions of their work.
As a consequence, one main point on which I differ from many other historians engaged in experimental replications is the question of whether it is allowable to make variations on the conditions of the original experiments. You will have seen that I was quite free in substituting modern materials and new procedures in place of the original in nearly all of my experiments. This would be anathema to many historians.
There are two modes of experimental work I have engaged in. First, I have tried to replicate the same phenomena as reported by the historical sceintists, for example in my replication of De Luc's work on boiling by slow heating (Experiment 5) and the boiling of de-gassed water (Experiment 6). In those experiments I took the modern volumetric flask as an unproblematic approximation to De Luc's long-necked flask, and I readily used a hotplate or a graphite bath instead of De Luc's oil bath for slow heating. In Experiment 6, I devised a different de-gassing procedure as I could not face De Luc's 4 weeks of shaking. Similarly, in Experiment 3, I used a modern method of creating a hydrophobic surface instead of Marcet's sulphur-coating. I believe that such variations are justified because I believe that they generate the same phenomena as what De Luc was observing. But does it even make sense to talk about the "same phenomena"? Yes, as much as it makes sense when other replicators talk about using the "same materials" or the "same conditions" as in the original experiments. It is going to be impossible to create exactly the same conditions that occurred in someone's experiment 200 years ago; we would not even know how to verify such complete replication, given that the descriptions we have are necessarily incomplete, and even instruments that have been preserved may have undergone material changes over time. These are simply limitations that we have to accept in any experimental replications.
Secondly, I was also interested in making extensions of the old experiments to explore other related phenomena. For example, in Experiment 2, having confirmed Gay-Lussac's report of the glass-metal difference, I extended my investigations to different types of metallic vessels, and to ceramic. Similarly, after observing the effects of traditional-type boiling chips, I observed the effects of the modern Teflon boiling stones in Experiment 4.
There are many different purposes in making experimental replications. For instance, Otto Sibum (1995) wanted to learn about the tacit dimension in experimental work. Peter Heering (1992) wanted to solve a historical puzzle, when he made a precise replication of Coulomb's torsion-balance experiment in electrostatics. Melvin Usselman et al. (2005) wanted to make an epistemological point about the nature of scientific knowledge. And I think it is important to have a pluralistic acceptance of all these various purposes. My own work in experimental replication serves the purposes of complementary science.
University College London - Gower Street - London - WC1E 6BT - +44 (0)20 7679 2000 - Copyright © 1999-2005 UCL