Dr Roger Griffin

Astronomer, Cambridge Observatories, UK

40 years of stellar spectroscopy

Roger Griffin


In 1969 I was working at the Cambridge Observatories, where I have been all my life since arriving as a graduate student there in 1957 - apart from one year when I had a postdoctorate position with the Carnegie Institution of Washington at what were then known as the Mount Wilson & Palomar Observatories.

My interests were in stellar spectra, and in the measurement of stellar radial velocities (velocities in the line of sight, determined by the Doppler shifts of the spectra). Such velocities had always been determined, since photography became useful in the late nineteenth century, by exposing a photographic plate to the star spectrum and also to the spectrum of a laboratory emission source through adjacent parts of the length of the same spectrographic entrance slit. After processing the plate one then had to measure the stellar and laboratory spectra line-by-line in a high-class measuring machine (travelling microscope).

In the 1960s I developed a new method here, at the on-site 36-inch reflecting telescope, whereby the stellar spectrum was cross-correlated in real time with a mask that looks like a high-contrast photographic negative of a typical star spectrum.

The spectrum is imaged on the mask, which is scanned in the wavelength coordinate, and at one position the transparent gaps in the mask are systematically occupied by the dark absorption lines in the star spectrum, so a minimum of light is transmitted, and the position where the minimum occurs is a measure of the wavelength shift. The transmitted light is continuously monitored by a photomultiplier as the scan takes place. The method was described in Astrophysical Journal, 148, 465, (1967).

But although it was very effective and far better in terms of sensitivity, precision and ease of use than the old method, I encountered a lot of resistance to it among the people who had spent their lives using the old method, and they were no doubt the referees who greatly obstructed all my efforts to publish papers giving results of the technique. Nobody else adopted it.

For ten years I made more and better measurements here than the rest of the astronomers in the world put together - with the one exception that the Palomar Observatory, which had not been interested in radial velocities previously, invited me and a collaborator, James Gunn, to make a second-generation instrument of the same sort for the 200-inch reflector which was then the largest telescope in the world. We duly made such an instrument (Astrophysical Journal, 191, 545, 1974), and it was very successful.

Eventually I decided that an advertising campaign was called for.  I had been measuring the radial velocities of binary stars; one can determine their orbits from systematic measurements of their velocities as they circulate (orbital periods range from less than a day up to far too long to watch). At the time I was one of four Editors of the journal 'The Observatory', which is published every two months.

Naturally, resistance to publication was less in that journal than in others, and I had the idea that I could start a series of papers giving orbits of binary stars, obtained by the new photoelectric method, and by including the words 'photoelectric radial velocities' in the series title I could flourish the method before the astronomical public in paper after paper. Having several years' data already 'in the bank', I felt that as long as my interest held up and the Editors would accept the papers I could put a paper in every issue of 'The Observatory' indefinitely!

In actual fact, only a few years after the series started, in 1975, the resistance collapsed and the astronomical world embraced the method itself, and long ago it became the case that nobody would dream of measuring velocities in any other way than by cross-correlation.

Almost all the evidence for such diverse things as black holes and extra-solar planets is derived by the method. If I got a citation every time anyone measured anything spectroscopically by cross-correlation my citation record would increase by hundreds a week! - but nobody pays me that compliment, any more than they make acknowledgement to Doppler. 

So the original purpose of the series of papers became otiose, and the series title has for a long time appeared tautological because *all* radial velocities are determined photoelectrically! But there seemed no purpose in stopping the series, which was documenting interesting binary-star orbits, and in fact it has continued ever since. A landmark was passed last summer, when Paper 200 was reached.

The series has documented substantially more orbits of cool stars than the *total* number that had ever been determined at the time that the series started, and of course other people too have contributed many orbits. The method has not only greatly rejuvenated the field in which I have been working myself but has also made possible all sorts of other observations, including the ones mentioned above, that previously were not thought of, and which, if thought of, could not have been pursued.

The original instrument, with which I developed the technique, was made on-site here, with critical elements of it made by me personally, and was rather in the Cambridge tradition of string and sealing wax (and the Palomar instrument incorporated 11 Meccano motors!). The novel part of it, in the vicinity of the focus of the spectrum, went to the Science Museum at the Museum's request as an historic instrument directly it was retired from active research work. Though now nominally retired myself, I am still observing with the same 36-inch telescope, though with of course a considerably improved spectrometer.