these velocity changes. So they need highly reliable local ‘standards of rest’ — standards of precisely known and completely stable wavelengths. Until now, discharge lamps and absorption spectra have served well. The first technique provides stabilized, reasonably well- distributed emission lines for elements such as thorium and argon (Fig. 1b); the second works by passing starlight through a cell containing a molecular vapour, usually iodine, so that the molecule’s absorption lines are imposed on it. With further improvements such as the use of a highly stable, preferably evacuated, spec- trograph, Doppler measurement precisions of about 1 m s –1 can be achieved using reference standards derived from either or both of these techniques. The harvest from the radial-velocity method has been, at the time of writing, a crop of 261 giant planets. Some of these are rocky, and 25 of the solar systems involved have been found to have more than one planet 2 . The push is now on to find analogues of Earth — rocky planets within the habitable zone (the range of distances within which liquid water would be present) of their parent stars. The focus will be on stars that are both cooler and less massive than the Sun: in these solar systems the radius of the habitable zone is considerably smaller and closer to the star than it is in our own, and a planet of lower mass causes a larger velocity perturbation of the star over a shorter time. Such cool stars are considerably brighter at infrared wavelengths than they are at optical wavelengths, and the intention is to build spec- trographs specifically designed to cover this region. But therein lies a problem: at infrared wavelengths, there are no suitable extended emission or absorption spectra for use as a reference standard. Enter the ‘astro-comb’. Li and colleagues’ brainchild 1 is an optically filtered comb of evenly spaced frequency references, all derived from a single frequency source — a pulsed laser. The idea is not new 3,4 (half of a Nobel Prize in Physics was awarded for the idea in 2005) 5 , and nor is its application to astronomy 6 . But Li et al. are the first to realize the concept in a way suitable for astronomical practice, in what could be a breakthrough in the precision of astronomical spectroscopy. The principal difficulty that the authors had to overcome was that, although the repetition rate of a fast pulsed laser generates a wide, sta- ble comb of equally spaced sharp fringes, this spacing is much too fine to be resolved by the planet-search spectrographs currently in use or proposed. Li et al. use a Fabry–Perot optical filter to give a much sparser comb with teeth ideally spaced for optimum wavelength cali- bration (Fig. 1c). This might sound simple, but to cover the possible range of frequencies to be measured an effective astro-comb needs to provide at least a thousand teeth, each acting as a reference standard for a particular part of the spectrum, and each bright enough to be readily exploited in a short space of time. That makes great technical demands on the avail- able equipment. In theory, the astro-comb could lead to radial-velocity measurements with a preci- sion of 1 cm s –1 — nearly a hundred times better than the current best, and comparable to the magnitude of Earth’s influence on the Sun. But the advance is only a first, albeit essential, step when it comes to the detection of terres- trial planets. There are many factors that can confound accurate measurements of the true radial velocity of a star: if our own Sun serves as a benchmark, a star’s surface is a maze of bright granules caused by hot gas boiling to the surface at velocities of around 1 km s –1 ; dark sunspots, coupled with the star’s rotation, severely distort the measured radial velocity; and rapid pulsations, a characteristic of most stars, further complicate matters. Extracting any signal corresponding to another Earth has substantial hurdles to overcome yet. ■ Gordon Walker is emeritus professor in the Department of Physics and Astronomy, University of British Columbia, 6224 Agricultural Road, Vancouver, British Columbia V6T 1Z1, Canada. e-mail: gordonwa@uvic.ca 1. Li, C.-H. et al. Nature 452, 610–612 (2008). 2. http://exoplanet.eu 3. Cundiff, S., Ye, J. & Hall, J. Sci. Am. 298 (4), 74–81 (2008). 4. Udem, Th., Holzwarth, R. & Hänsch, T. W. Nature 416, 233–237 (2002). 5. http://nobelprize.org/nobel_prizes/physics/ laureates/2005/index.html 6. Murphy, M. T. et al. Mon. Not. R. Astron. Soc. 380, 839–847 (2007). Over the past decade, the word ‘hum’ has acquired a special meaning for seismologists. No longer just what they might do under the shower, it connotes for them a fundamental resonant oscillation of the Earth. A sequence of these oscillation modes, with periods of between around 2 and 5 minutes, was first identified 1–3 in 1998. These were all ‘spheroidal’ modes, representing perturbations of the planet’s equilibrium surface, rather akin to the effect of waves on water. Writing in Geophysical Research Letters 4 , Kurrle and Widmer-Schnid- rig now introduce a further, entirely different mode — ‘toroidal’ hum, in which parts of Earth’s surface twist around in the horizontal plane (Fig. 1, overleaf). The existence of this low-frequency Earth hum is not the surprising thing. Seismic noise is ubiquitous, generated by various natural processes such as falling water (the impact of, say, the Niagara Falls is not confined to the surface) and even swaying trees, as well as all manner of human activities. It is the magnitude of the hum that is disconcerting 5,6 : its summed amplitude is equivalent to a continuous earth- quake of magnitude 6. (Because the waves are at such a low frequency, we humans cannot sense them; as they represent no threat to our well-being, there has presumably never been a need to evolve such a capability.) An earth- quake of this size occurs once every three days on average; clearly, seismic activity cannot sus- tain hum of such magnitude and continuity. Since those first intriguing findings, the ocean has by general consensus been identi- fied as the most likely source of Earth hum: the origin of the excitations seems to lie in oceanic areas at mid-latitudes, between about 30° and 60° north and south 7,8 . In addition, the amplitude of the effect has a periodicity of six months, with a maximum occurring in each hemisphere during its winter; satellite data show that ocean waves are particularly large at mid-latitudes during the winter months. The proposal 9,10 , which borrows an idea of some 60 years ago 11 , is that so-called infra- gravity waves, which are known to have the same sort of periods as the hum 8,12 , transmit this oceanic motion to the solid Earth. These waves are similar to tsunami waves — low- frequency, long-wavelength ocean waves that move the whole column of ocean waters, from surface to sea floor, as they propagate. The col- lision of such waves could produce large pres- sure variations 11 , and thus excite the hum. A problem is that infragravity waves are mainly known to be a phenomenon of shallow water, although a mechanism for generating them in the deep ocean has recently been proposed 13 . Even so, a direct interaction between the atmosphere and the solid Earth has not been ruled out as a source of the hum. Atmospheric and oceanic effects are difficult to separate: when we see large-amplitude ocean waves, the cause is likely to be an atmospheric effect, namely strong winds. The observant frequent flyer from New York to Paris or Tokyo to San Francisco will note that, during winter in the Northern Hemisphere, flights are often diverted from the shortest geographical route, a great circle over the Arctic, to a more south- erly route of near-constant mid-latitude. The GEOPHYSICS Humming a different tune Toshiro Tanimoto Earth breathes in and out, murmuring gently to itself as it does so. The habit has been ascribed to the tickling effects of ocean waves — but a new-found twisting oscillation might reopen the search for the source. 539 NATURE|Vol 452|3 April 2008 NEWS & VIEWS