Introduction

Any of us who has used the Global Positioning System (GPS) in one of the gadgets of everyday life has also relied on the accuracy of the predictions of Einstein's theory of gravity, General Relativity (GR). GPS systems accurately provide your position relative to satellites positioned thousands of kilometres from the Earth, and their ability to do so requires being able to understand time and position measurements to better than 1 part in 1010. Such an accuracy is comparable to the predicted relativistic effects for such measurements in the Earth's gravitational field, which are of order GM⊕/R⊕c2 ~ 10−10, where G is Newton's constant, M⊕ and R⊕ are the Earth's mass and mean radius, and c is the speed of light. GR also does well when compared with other precise measurements within the solar system, as well as in some extra-solar settings.

So we live in an age when engineers must know about General Relativity in order to understand why some their instruments work so accurately. And yet we also are often told there is a crisis in reconciling GR with quantum mechanics, with the size of quantum effects being said to be infinite (or – what is the same – to be unpredictable) for gravitating systems.

Introduction

How does a proposal for unification go from an interesting body of mathematical results to a plausible explanation of natural phenomena? While evidence of mathematical consistency is ultimately important, what is often decisive is that a proposed unification leads to predictions of phenomena that are both new and generic. By generic I mean that the new phenomena are general consequences of the proposed unification and thus hold for a wide range of parameters as well as for generic initial conditions. The proposal becomes an explanation when some of those new generic phenonena are observed.

Generic consequences of unification often involve processes in which the things unified transform into each other. For example, electromagnetic waves are a generic consequence of unifying electricity and magnetism, weak vector bosons are a generic consequence of unifying the weak and electromagnetic interactions, and light bending is a generic consequence of the equivalence principle which unifies gravity and inertia.

Looking at history, we see that the reasons why proposals for unification succeed or fail often have to do with their generic consequences. In successful cases the consequences do not conflict with previous experiments but are easily confirmed when looked for in new experiments. These are cases in which we come to celebrate the unification. In bad cases the consequences generically disagree with experiment. Some of these cases still survive for some time because the theory has parameters that can be tuned to hide the consequences of the unification.

SPECTROSCOPY OF THE SOLAR CHROMOSPHERE

The discovery of helium

A new era in solar spectroscopy was launched in the late 1860s when a series of solar eclipses provided an opportunity to study the solar chromosphere. This is the hot tenuous layer lying above the photosphere, which is the region of the Sun that emits the vast proportion of the visible light. At the time of a total solar eclipse, it was well known by the 1860s that the light from the thin chromospheric layer (with a height of ∼ 10 to 12 arc seconds or 8000 to 10 000 km) became briefly visible. Moreover, the solar prominences are large structures of chromospheric material extending out from the limb, often a minute of arc or more in height. They too were seen at times of eclipse, and the question of the physical conditions in the chromosphere and prominences arose. The spectroscope was the natural tool to settle the issue. If they were comprised of hot low density gas, then bright emission lines would be expected, as William Huggins pointed out, and hence they should have spectra similar to those of the gaseous nebulae, such as the Orion nebula. Huggins also believed that if bright lines were present, then perhaps they could be observed with suitable glass filters to isolate the line radiation, even outside of eclipse, but attempts to do so did not come to fruition.

Few astronomers would dispute the pivotal rôle that the astronomical spectrograph has played in the development of astrophysics. Of all astronomical instruments other than the telescope itself, none other can compete with the spectrograph for the range of new astronomical knowledge it has provided, and for the insights it has given on the physical nature of the celestial bodies in the Universe. Together with the predecessor of the spectrograph, the visual spectroscope, these instruments have revolutionized our knowledge of the Sun, the planets, stars, gaseous nebulae, the interstellar medium, galaxies and quasars.

Without the spectrograph, we would know nothing of solar or stellar composition, nothing about stellar rotation rates, and much less than we do on stellar space motions and binary stars. Even the real nature of the stars themselves would be a matter of conjecture and debate. And we would have rudimentary knowledge of the conditions prevailing in gaseous and planetary nebulae and of the nature of external galaxies beyond the Milky Way. There would be no Hubble's law, and hence no direct knowledge of the expansion of the Universe other than indirect inference based on Olbers' paradox or on theoretical prediction. Quasars would not be easily distinguished from stars, and the study of radio galaxies and active galactic nuclei would be limited to their morphological properties in optical or radio images. In short, optical spectrographs have underpinned almost every branch of astrophysics in the past century and a half.

SPECTROSCOPY OF THE SOLAR CHROMOSPHERE

The discovery of helium

A new era in solar spectroscopy was launched in the late 1860s when a series of solar eclipses provided an opportunity to study the solar chromosphere. This is the hot tenuous layer lying above the photosphere, which is the region of the Sun that emits the vast proportion of the visible light. At the time of a total solar eclipse, it was well known by the 1860s that the light from the thin chromospheric layer (with a height of ∼ 10 to 12 arc seconds or 8000 to 10 000 km) became briefly visible. Moreover, the solar prominences are large structures of chromospheric material extending out from the limb, often a minute of arc or more in height. They too were seen at times of eclipse, and the question of the physical conditions in the chromosphere and prominences arose. The spectroscope was the natural tool to settle the issue. If they were comprised of hot low density gas, then bright emission lines would be expected, as William Huggins pointed out, and hence they should have spectra similar to those of the gaseous nebulae, such as the Orion nebula. Huggins also believed that if bright lines were present, then perhaps they could be observed with suitable glass filters to isolate the line radiation, even outside of eclipse, but attempts to do so did not come to fruition.

ULTRAVIOLET AND NEBULAR SPECTROSCOPY

Henry Draper and William Huggins, pioneers in ultraviolet stellar spectroscopy

Observational studies of the near ultraviolet region of stellar spectra have a long history, which goes back to the early days of stellar spectrum photography. The very first spectrum ever recorded by photography was by Henry Draper in 1872. He used his 28-inch reflector and a spectrograph with a quartz prism, and the then relatively new innovation of a dry emulsion glass plate. He noted:

The key to Draper's success was in part his use of the new dry plates, which were so much more convenient than the wet collodion plates used previously in astronomical photography. But also his use of a silvered-glass reflecting telescope and a spectrograph with a quartz prism allowed him not only to go below the approximately 400 nm wavelength limit of the human eye, but below the approximately 380 nm limit for the transmission of flint glass used in the lenses of achromatic refractors.

GENERAL PROPERTIES OF A SPECTROGRAPH

A spectrograph is an instrument that receives light from a source, disperses the light according to its wavelength into a spectrum, and focusses the spectrum onto a detector, which records the spectral image. In the astronomical case the source might be a star or galaxy, and the light will first be collected by a telescope. Many telescopes produce an image of the source on a spectrograph slit (although slitless objective prism instruments are also possible). After the slit, a collimator renders the rays almost parallel, and a dispersing element, usually a grating or a prism, sends photons of different wavelengths into different directions. A camera then records a continuous succession of monochromatic slit images, each displaced in the dispersion direction according to its wavelength. This array of slit images constitutes the spectrum.

The simplest possible slit spectrograph (Fig. 2.1) therefore comprises a slit, a collimator (either a mirror or a lens system), a dispersing element (typically a grating or a prism) and a camera (again a mirror or lens system) and finally a detector (perhaps a charge-coupled device or CCD, perhaps a photographic plate, but in early instruments it was the human eye in conjunction with an eyepiece).

INTRODUCTION TO SPECTROGRAPHS OF THE LATE TWENTIETH CENTURY

In this chapter, ten outstanding spectrographs which had their first light in the late twentieth or early twenty-first centuries are presented. The section covers five high resolution instruments, four at low resolution and one spectrograph with multiple capabilities that includes low, medium and high resolution modes. All were commissioned between 1984 and 2003 and they are presented in chronological order of their entry into service.

The choice of ten representative spectrographs from the several dozen commissioned over this two decade period was a matter of some difficulty. The final choice was my own personal one. I wanted to show that spectrograph design has become a pursuit of great innovation, ingenuity and also complexity, at times verging towards a creative art-form involving cutting edge optical technology. In this period, huge developments were made in the use of optical fibres, larger and improved efficiency CCD detectors, the production of large mosaic gratings, the development of grisms and volume phase holographic gratings, the design and manufacture of multi-element dioptric cameras with specialized antireflection coatings, high reflection coatings on mirrors, and the extension of the wavelength range both down into the near ultraviolet at the atmospheric limit and also to the far red near one micrometre.

LOW RESOLUTION MULTI – OBJECT SPECTROSCOPY

For almost a century since the objective prism spectrograph was first used at Harvard in 1885, this was the only solution available to astronomers for multi-object spectroscopy – the simultaneous recording of the spectra of more than one object. All other astronomical spectroscopy was undertaken sequentially, one object at a time. The limitations of the objective prism caused by overlapping spectra and sky brightness, as well as variable resolving power (dependent on the seeing) have already been discussed.

From about 1980, two new solutions for multi-object spectroscopy, which overcame some of the problems of the objective prism, were introduced. One is the aperture plate combined with a low dispersion element, which is often a combined prism and transmission grating (or ‘grism’ – see Section 2.8). The other is the use of optical fibres. The early history (roughly the first decade) of both these techniques is described here.

APERTURE PLATE MULTI – OBJECT SPECTROSCOPY

The aperture plate was introduced by Harvey Butcher (b. 1947) at Kitt Peak in 1980 as a way of undertaking multi-object spectroscopy of faint objects, all of which lay within the 5 arc minute field of view at the Cassegrain focus of the 4-m Mayall telescope. The aperture plate is a black anodized aluminium sheet in the focal plane, with holes drilled in it, typically with a diameter corresponding to 2.5 arc seconds on the sky.

COLOUR, REFRANGIBILITY AND WAVELENGTH

The concepts of colour, refrangibility and wavelength have been crucial for the development of spectroscopes and spectroscopy. Indeed the first prism spectroscopes were built partly for measuring the refractive indices or refrangibility of different glasses, while diffraction gratings were used for early measurements of wavelength.

The refraction of light

The Dutch astronomer Willebrord Snell (1591–1626) is usually credited with the discovery of the law of refraction, in the early seventeenth century in Leiden. René Descartes (1596–1650) later included the law in his treatise Dioptrics of 1637 (without however acknowledging Snell), and he used it to account for the formation of a rainbow, but not explicitly for the colours that the rainbow produces.

Isaac Newton's (1642–1726) first paper, published by the Royal Society in 1672, but based on his optical experiments undertaken six years earlier, came to important conclusions on the relationship between refrangibility and colour. Newton studied the refraction of the Sun's rays in a glass prism, and projected the resulting spectrum onto a wall of his room. He concluded: ‘… light consists of rays differently refrangible, which, without any respect to a difference in their incidence, were, according to their degrees of refrangibility, transmitted towards divers parts of the wall.’ He found that the Sun's light was composed of rays of different primary colours, each with its different refrangibility, which gave rise to the dispersive power of the prism.