The experimentalist’s skill

Some scientists have an unusual skill for extracting precise data from their instruments.  Sometimes this leads others to question their results.

A few weeks ago our navigator was reminded, again, that operating a sextant takes some learning.  It’s not learning in the same way that one learns a bit of mathematics, or a foreign language.  It’s less objective and more a matter of skill: how to hold the instrument steady, of course, but also how to see the picture clearly and extract a good reading.  Most mariners can do a good job of it; some are very good indeed.

As we’ve mentioned, nineteenth-century observational astronomy used the same kind of technique: seeing small details to make precise measurements and timings.  And it required a similar development of personal skill.  For those who measured double stars (by placing fine crosshairs on them and reading off micrometer dials), it was a rule of thumb that a new observer required at least a year of work before his data could be used.  And even then, among those who used these measurements to work out the orbits of the stars, it was known whose data were more reliable and whose might be ignored.

Indeed, the experimental side of physical science retains much of this character.  The descriptions in textbooks of how things were discovered or measured are necessarily simplified, to show the principles.  But to set up an experiment, track down all the various problems, and extract very precise data is indeed a skill.  It goes far beyond just reading a dial.  That means that, on the very edge of research, there may be only a very few who can actually do the measurement or detect the effect.

Especially for those who are not working in the field this raises doubts.  Shouldn’t science be reproducible?  It’s even led to the claim that science is nothing more than what is agreed among some self-appointed elite.

Well, Royal Astronomical Society certainly did not decide what double-star orbits should be and then order its observations accordingly.  But it has happened that some cutting-edge results turned out not to be true.  What saves us most of the time is improvements in technology, which allow more precision to everyone; or at least allow more repetitions of a given experiment.  Eventually an error can be identified as such.  In addition, nowadays much of the data are recorded automatically, by devices whose characteristics are well known (rather than by humans, who exercise judgement).

But to make a cutting-edge experiment or observation in the first place you still need someone with a certain experimentalist’s skill.