How science is done (I)
We look at an extraordinary data set.
A recent article in an astronomy journal is unusual, and is of a kind that is becoming more so. It is based on a collection of 559 orbits of binary stars, stars orbiting around each other. Binaries are not unusual, nor are papers analyzing them. In fact stars in binary (and multiple) systems outnumber lone stars like the Sun. Sometimes the orbit can be traced on the sky, though stars far enough apart to be seen separately generally have very long orbits (decades to centuries) and thus reveal their orbits only slowly. These 559 were detected through Doppler shifts of their spectral lines, which is also not unusual. But the observations on which the orbits were based were all collected by one man (with collaborators at times) over a 45-year program; and that is unusual.
To get an orbit Roger Griffin first has to have some reason for thinking that a star is binary. Often this comes about by searching published redshifts and noting any discrepancies, that is, figures that differ by more than is reasonable given the equipment involved. Mostly they come from his own survey program, checking redshifts once and then again after an interval. Once placed on his list he will re-observe a star as often as its behavior seems to indicate. For each star, the scheduling of re-observing is different. Orbital periods of his 559 stars vary from 0.31 days to 31292 days (about 86 years). Very eccentric orbits might show little or no variation except for a short period around perihelion, which has to be captured in detail for a good result. A binary with a period close to an even multiple of an Earth year, and well-placed in the sky for observation for only a short time during the year, will be caught at nearly the same place in its orbit each time; forcing Roger try to catch it close to dawn or sunset, or patiently observing again over many years.
The trend in modern astronomy, as we’ve noted, is toward large collaborations using many pieces of expensive gear. Surveys are automated, with data recorded at a rate such that no human can even glance at it all. The highest-profile observations are made with equipment that could not have been built a decade ago. And the lifetime of a research grant is typically 3-5 years; if there aren’t plenty of publishable results in that time, a renewal is unlikely.
Roger’s program, using essentially the same technique for 45 years, manually scheduled and carried out, is thus highly unusual. (It has also largely been done with a 1-meter telescope, tiny by modern standards, at a poor site: Cambridge, UK. Not only is the weather often uncooperative, but the University authorities years ago decided that well-lit sports fields were more important than dark skies for telescopes.)
None of Roger’s papers reporting his results would be considered breakthroughs, of the type sought by the highest-visibility journals. But his data are invaluable in many ways to astronomers working on stellar structure and evolution.
So will Roger be superseded by some modern version of the project? It would not take much investment, as astronomy goes nowadays, to set up 1-meter ‘scopes with his kind of instrument at good sites in the northern and southern hemisphere. It would be much harder, we think, to automate the target selection and scheduling tasks, and to secure funding for such a long term. Perhaps hardest to automate would be Roger’s way of evaluating his data, about which we’ll speak next time.
