The Hubble Space Telescope USEFUL IN SPITE OF THE MIRROR G. Meylan Space Telescope Science Institute Baltimore, Maryland, USA Assessments of the Hubble Space spread function, i.e., the image intensity Telescope (HST) and comparisons with distribution of a point source, possesses a ground-based observations of several sharp central core over a large range of well-studied astronomical objects de­ secondary mirror positions. monstrate the instrument's unique ob­ ESA's Faint Object Camera consists of servational capabilities. This is in spite of two independent optical relays, with a by now commonly recognised flaw. effective focal ratios (focal numbers) of Soon after launch in 1990, in-orbit testing f/48 and f/96, each of them being equip­ of the HST revealed a serious optical pro­ ped with a photon counting detector blem with the 2.4 metre primary mirror. capable of producing images using a va­ Prompted during the first two months of riety of filters and prisms. Fig. 1 shows a use by the consistent failure to align and FOC image of two stars located 10 arcmin focus the two (primary and secondary) East of ε Scorpii obtained in the f/96 mode mirrors of the telescope, a series of through a filter with a wavelength range images was taken with the secondary mir­ Δλ = 34 Å centred on a wavelength λ = ror stepped through the full range of posi­ 4870 Â. The fainter star has an intensity tions on each side of the nominal focus. V in the visible wavelengths centred on These data showed conclusively that the approximately 5500 Â which is a factor of anticipated sharp focus could not be ob­ about 4.5 times smaller than V for the tained at any position of the secondary brighter star : the separation between the mirror. Moreover, images did not show the two stars is 6.3 arcsec. symmetric behaviour with respect to ne­ Although the exact focus position to be gative and positive offsets around the used may still change, the image can be expected best focus position. It was con­ regarded as representative of the present cluded that the telescope suffers from a point spread function with its characteris­ substantial amount of spherical aberration tic features, namely a sharp central bright owing to an error introduced when the core (Fig. 1, star on the left) and an exten­ primary mirror was ground more than a ded faint halo (Fig. 1, star on the right). decade ago. Several diffraction rings can be identified A high angular resolution and the possi­ in the outer parts of the image of the bility to observe in the ultraviolet remain bright star on the right : the larger outer the telescope's main advantages. Its inhe­ ring structure is due to the shadow of the rent problem and important potential will secondary mirror and the radial "tendrils" be illustrated using a few observations Fig. 1 — Image (in negative) of two stars reflect the diffraction pattern of the se­ made with the Faint Object Camera (FOC), located 10 arcmin East of ε Scorpii obtained condary mirror support spider and moun­ one of the European Space Agency's with the Faint Object Camera in the f/96 ting "pads" located on the primary mirror. (ESA) contributions to the HST project. mode through a filter with Δλ - 34 A cen­ While the light from the star is spread Less Photons — But Still Hawk-Eyed tred on λ = 4870 A : the brighter star has a over a circle with a diameter of almost 5 The spherical aberration discovered in visible magnitude V = 12.3; the second arcsec, the central peak of the image is star is about 4.5 magnitudes fainter. The nonetheless extremely sharp, even though the HST optics gives rise to grossly defo- separation between the two stars is 6.3 arc- cused images in the telescope's focal it contains only 15% of the energy instead sec. The bright star shows the central core of the expected 80%. The loss of energy plane at all settings of the secondary mir­ surrounded by a large halo; the faint star ror. However, because one anulus of the is clearly visible in Fig. 1b where the simu­ has no halo, only its core emerges out of the lated, originally expected, point spread primary mirror is always in focus at all background noise. settings between the foci of the outer and function (left) calculated for a point b, middle). The simulated, originally expec­ source with allowance for instrumental inner edges of the primary mirror (margi­ ted, point spread function (left) with the nal and paraxial foci), the telescope point corresponding observed point spread func­ effects is presented, together with the tion (right) obtained in orbit through the observed function (right) obtained in orbit Georges Meylan has been a Staff Astronomer same filter. The actual core contains about through the same filter for the bright star with the Space Telescope Science Institute, 15% of the energy instead of the 80% on the left of Fig. 1a. The intensity reduc­ 3700 San Martin Drive, Baltimore, MD 21218, expected, corresponding to a loss in sensi­ tion is dramatic, corresponding to about USA, since 1989. He studied mathematics at tivity of about three orders of magnitude. three orders of magnitude. However, Fig. the University of Lausanne, Switzerland, and 1c demonstrates that, apart from some astrophysics at the University of Geneva, re­ c, lower) The same data as b) but with a dif­ ceiving a Ph.D in 1985. He then held postdoc ferent vertical scale : the sharp core is still energy, everything has not been lost. The positions at the University of California, Berke­ present in the observed image so for bright same simulated (left) and observed (right) ley, and at the Max-Planck-Institut, Garching, objects, the resolving power of the tele­ point spread functions are displayed, but Germany. scope is retained. with a different vertical scale. A sharp 206 Europhys. News 22 (1991) central core is still clearly present In the Microlensing observed data and its diameter is only The photograph is ideal for accurately three pixels wide, corresponding to 0.066 measuring the positions and brightness of arcsec at the full width half maximum the four quasar images — brightnesses (FWHM) intensity. which are already known to vary because Thus, owing to the flaw In the primary of microlensing, when not only the whole mirror, the point spread function of the potential of the galaxy is involved but also FOC has a tight diffraction-limited core, as individual stars in their motions near the expected originally, surrounded by an ex­ line of sight to the lensed quasar. This tended halo due to the spherical aberra­ happens only when the lensing galaxy is tion. It Is not so much the high spatial relatively close to the observer. The (angu­ resolution of the HST cameras that has lar) proper motions of the stars are then been compromised by the error in the fast enough to produce variations of the mirror, but the limiting sensitivity. intensity of the images, hopefully on It is worth emphasising that the astro- human time scales, i.e., over some weeks, physical observations described below months or a few years at the most. Such were made essentially to assess the conditions are met here owing to the HST's present capabilities. Consequently, unusually close proximity of the lensing they were done only for objects allowing galaxy, and variations in the relative in­ comparison, i.e., with those already been tensities of the four images are observed. extensively observed with the most ad­ We have found that the relative intensi­ vanced ground-based techniques. ties have varied randomly over the last two years, e.g., as shown in Fig. 2, the A An Ideal Case of Gravitational Lensing : component, formerly the brightest, is now The Einstein Cross fainter than component B — an effect Since the first theoretical discussions caused by stars in the galaxy passing more than 70 years ago on the phenome­ across the line-of-sight. non of light rays bent by intervening mass Thus, owing to the unusual position of in the Universe — Eddington in 1920 for the lensing galaxy, the large number of the case of the Sun, Einstein in 1936 for Fig. 2 — The Einstein Cross, or G2237 + images produced, the relatively small time stars, and Zwicky in 1937 for galaxies — 0305, taken by the Hubble Space Telescope delay between images, and the likelihood gravitational lensing has steadily grown with ESA's Faint Object Camera in the f/96 of frequent microlensing events, the to become one of the most active fields mode through a filter centred on a wave- G 2237 system represents an excellent of research in extragalactic astronomy. length λ = 4850 A with an exposure time candidate for monitoring lensing beha­ There have been numerous theoretical of 25 minutes : a, upper) The raw image viour and for determining independent investigations, but the number of candi­ showing that the single image of the quasar estimates of fundamental astrophysical dates for which a body of evidence sup­ is split into (in this particular case) four parameters such has the Hubble constant ports true lensing effects remains small. images surrounding the nucleus of the ga­ H0 , a cosmological parameter linked to For it is only during the last decade that a laxy. The nucleus of the intervening galaxy the age of the Universe. few quasar systems have been found to lies at only about 0.75 arcsec from the be in reasonable agreement with the gra­ quasar images. No Massive Star at the Core of vitational lensing interpretation.