The Machine Speaks Falsely

How can one determine if an experimental apparatus is giving an incorrect result, if it is speaking falsely? An interesting example of this occurred in the experimental investigation, in the early twentieth century, of the energy spectrum of electrons emitted in β decay. Meitner and her collaborators (1911), using photographic detection, found that all the electrons emitted by a single radioactive element were monoenergetic. Chadwick (1914), on the other hand, using either an ionization chamber or a Geiger counter, found a continuous energy spectrum. Meitner et al. proposed various mechanisms whereby initially monoenergetic electrons might lose energy. These were shown to be unsatisfactory, although the possibility of an unknown mechanism for energy loss remained. In 1927 Ellis and Wooster, using a total-absorption calorimeter, which eliminated all of these possibilities, demonstrated that the energy spectrum was indeed continuous. It had taken fifteen years to show that the photographic detection had spoken falsely.