peak was expected to be a challenge for the field for several years to come. Instead, the three groups reporting in this issue 1–3 have found a new physical regime, in which electrons are ‘self-injected’ in a narrow region of space and made to surf as a single group, all reaching the same energy (Fig. 1). The three experiments are similar in many ways.In each of them, 10–30 terrawatts of laser power, in pulses 30–55 femtoseconds long, is focused into an ion- ized jet of gas roughly 2 mm long and with a particle density of 2ǂ10 19 cm ǁ3 ; a nearly monoenergetic distribution of electrons is observed, with instrument-limited energy spreads of 2–24% at roughly 80–170 MeV. With up to a few times 10 9 electrons per beam, the energy densities in these experi- ments are a hundred to a thousand times higher than has previously been achieved. The angular spread of the beams is also about ten times tighter than before — com- parable to the best of the beams produced by radio-frequency systems. Moreover, the pulse lengths of the beams are about 10 femtoseconds (10 ǁ14 s), making them attractive as potential radiation sources for ultrafast time-resolved studies in biology and physics. Despite the similarities between the three experiments, it is the differences that have helped to identify the mechanisms responsible for their success. The three groups used different approaches to control what turns out to be a key factor — the inter- action length in the plasma. The interaction length is the distance over which the parti- cles surf the wake, and it is determined by either the end of the plasma or the weaken- ing of the laser pulse through diffraction (the natural tendency of tightly focused light to spread). Geddes et al. 1 used a pre- formed plasma channel to guide the laser over several times the length that it would travel without diffraction in a vacuum; the groups of Mangles 2 and Faure 3 used a larger laser spot size (up to 24 micrometres) to increase the interaction length. The groups describe essentially the same physics: first, the laser pulse evolves to become shorter and narrower; this creates a large wake that The status of rice blast as a model system for studying aerial plant infection is based on its continuing impact on world food pro- duction, its amenability to molecular and genetic analyses, and the well-defined devel- opmental pathways it uses to invade aerial rice tissues. Infection occurs when airborne spores land on rice plants, sense the waxy aerial plant surface, and develop a dome- shaped organ called an appressorium. This produces phenomenal pressures — equiva- lent to those experienced in a deep-sea dive to 2,500 feet — enabling it to push a penetra- tion peg through the tough surface layers that protect the plant 2 . For the next week, the fungus spreads within the plant tissue and forms eyespot-shaped lesions (Fig. 1a), producing thousands of spores daily to invade new tissue. news and views 516 NATURE | VOL 431 | 30 SEPTEMBER 2004 | www.nature.com/nature traps electrons from the plasma; the loading of the wake with trapped particles turns off further trapping; and finally,‘dephasing’ of the electrons as they outrun the wake creates a monoenergetic beam (basically, like marbles that roll to the bottom of a hill, they arrive at different times but end up at the same energy; Fig. 1). Geddes et al. 1 emphasize the need for large interaction lengths to enable the electrons to dephase from the wave; their demonstra- tion of guiding an intense laser in a plasma channel suggests a means of extending future wakefield accelerators beyond the millimetre scale. Mangles et al. 2 , however, stress the need to reduce the interaction length to prevent the dephasing from becoming complete (the marbles reach the next hill and begin to slow down). Thus, as in the children’s story Goldilocks and the Three Bears, the interaction length must be not too long, nor too short, but just right. There is still a long way to go from these experiments in the 100-MeV range to the frontiers of high-energy physics (it’s likely that considerably more than 100,000 MeV needs to be available in a particle collision to produce a Higgs boson). The shot-to-shot stability and efficiency of these schemes also need to be improved. Nevertheless, these results represent the most significant step so far for laser-based accelerators, and should stimulate further advances in the near future. In particular, developments in high- power laser technology and plasma-channel production (particularly lower-density channels to increase the wake speed and hence the dephasing length) could both lead to the generation of beams of up to a few thousand MeV from a single-stage table-top device. Such accelerators would not only be more compact but would also exceed con- ventional sources in peak current, brightness and shortness of pulse duration. Wakefield acceleration may one day change the way we approach the physics and applications of particle beams. ■ Thomas Katsouleas is in the School of Engineering, University of Southern California, 3737 Watt Way, Los Angeles, California 90089-0271, USA. e-mail: katsoule@usc.edu 1. Geddes, C. G. R. et al. Nature 431, 538–541 (2004). 2. Mangles, S. P. D. et al. Nature 431, 535–538 (2004). 3. Faure, J. et al. Nature 431, 541–544 (2004). 4. Joshi, C. & Katsouleas, T. Physics Today 56, No. 6, 47–51 (2003). 5. Modena, A. et al. Nature 377, 606–608 (1995). 6. Malka, V. et al. Science 298, 1596–1600 (2002). Plant disease Underground life for rice foe Barbara Valent We still have much to learn about the world’s chief disease of rice — rice blast. That’s clear from the finding that the culprit not only infects aerial plant tissues but can also invade roots like a typical root pathogen. T he diverse fungi that threaten the world’s food crops are generally div- ided into those that infect plant struc- tures above the ground and those that infect roots. Fungi that attack aerial plant struc- tures use a few characteristic developmental pathways, and root-invading fungi — including symbiotic species that can be beneficial to plants — use different develop- mental routes. The rice blast fungus, which causes an annual loss of hundreds of millions of tonnes of rice worldwide, has become a model system for studying the aerial attack pathway. But, in a ground-breaking report that bridges the divide between the patho- genic lifestyles, Sesma and Osbourn 1 show that the foliar blast pathogen also invades roots, using a typical root-specific pathway (page 582 of this issue). Figure 1 Wakefield acceleration. a, In a plasma excited by a laser pulse, the wake potential rises until it steepens and breaks. Electrons from the plasma are caught in the ‘whitewater’ and surf the wave. b, The load of the electrons deforms the wake, stopping further trapping of electrons from the plasma. c, As the electrons surf to the bottom of the wake potential, they each arrive bearing a similar amount of energy. a b c 'Whitewater' of plasma electrons Plasma wake potential Loaded wake Mono- energetic beam Surfing electrons Laser pulse ©2004 Nature Publishing Group