Galileo and on

History helps to demonstrate why experiment has come to be important. The "scientific revolution" took place during the 16th and 17th centuries: explanations of nature derived from Aristotle's philosophy gave way to explanations based on a view of the world as a huge mechanistic system. Experiment was part of that revolution, and also its beneficiary. Modern experiment depends strongly on the notion that there is a knowable mechanism linking cause to effect. Experiments work by exerting control over a cause and noting the effects.

Of the heroes of the scientific revolution, the one most closely associated with experiment is Galileo Galilei. In "The Two New Sciences", his book on the nature of motion, he insist that experiment can free modern scientists from the ideas of the ancients, including Aristotle:

I seriously doubt that Aristotle ever tested whether it is true that (as he says) two stones, one ten times as heavy as the other, both released at the same instant to fall from a height, say of 100 braccia, differ so much in their speeds that upon arrival of the larger stone upon the ground, the other would be found to have descended no more than ten braccia … But I … who have made the test, assure you that a cannonball that weighs 100 pounds or 200 or even more does not anticipate by even one span the arrival on the ground of a musket ball of half (a pound), both coming from a height of 200 braccia.

So much for Aristotle; but to see Galileo's work as a simple triumph for empiricism is to miss some of the point. Galileo performed many experiments, and his accounts of them helped convince others of his ideas. But historians "seriously doubt" that his experiments provided results anything like as neat as the ones he described. For Galileo, "experiment" meant something like a trial, a trial of an idea against the world. But Galileo, like almost all experimenters, was not always a disinterested judge. He was an advocate, taking the testimony of the trial and using it to his own ends.

Experiments are almost always used as part of an argument. Because of that, there is an inescapable subjectivity to them - though that subjectivity need not and should not stretch to fraud. They are also, though, the cornerstone of objective knowledge. The way that scientists revolve this paradox is through the notion of replicability. A "good" experiments is one that, when repeated by someone else, produces the same result. The identity of the experimenter is thus made into an arbitraty detail.

The only other 17th-centry experimenter to rival Galileo in fame can be called on to prove the point. Robert Boyle's air - pump is one of the earliest forerunners of the complicated and temperamental machines that fill up today's laboratories. Boyle used it to develop a theory of pressure and vacuum - one which not every - one approved of. Thomas Hobbes, now remembered as a political philosopher but at the time thought of as a natural philosopher in the same sort of class as Descartes, was one of those who argued against Boyle's interpretation of the air - pump experiments. One of the ways that Boyle won the argument was by showing that other people, with other air - pumps, could achieve the same effects. Another of his strategies was to write extremely detailed descriptions of his work, which made the reader feel that, although he had notbeen there, he had seen everything that mattered and would be albe to reproduce the experiment if the need arose.

The idea of replicability still shapes the way in which scientific papers are written. Typically, an experimental paper comes in four parts: an introduction, a description of the method by which the experiment was done, the results, and a discussion of the results. It is like a subjectivity sandwich, with the theoretical ideas and conclusions wrapped around the chopped liver of reproducible data and methodology. In principle, anyone who reads the section on method will be in a position to achieve the same results.

In practice, though, experiment and theory are not so simply disentangled. The interpretation of an experiment depends on theory-normally the very theory the experiment is testing. Two people can do experiments which, to an outside observer, may look similar. They nay get results which the observer finds industinguishable. But if they start from different theoretical assumptions they need not end up agreeing.

Take the example of Antoine Lavoisier and Joseph Priestley. In the 1770s they both performed experiments which involved heating an ore of mercury and thus changing the nature of the air above it. Lavoisier saw the mercury ore - which he later came to call an oxide - giving off a new substance, which he called oxygen. Priestley saw a substance called phlogiston leaving the air and entering the ore, leaving the air above it "dephlogisticated". Dephlogisticated air had different properties from those of normal air; things burnt well in it, for example.

The two men, able experimenters with similar instruments, held different views of the world; between such different views, experiment cannot decide. Philosophers and sociologists of science talk of all observations being "theory laden": the theory you hold determines the language of observation, though such a goal can be worked towards, and what is achieved may serve pretty well as long as no one with a radically different view of the wold starts picking holes in it. Without support from firm slices of theorizing, the chopped liver of objectivity is too sloppy to handle.