Lasers are often described in terms of a light field circulating in an optical resonator system. Now a laser has been demonstrated in which the field resides primarily in the atomic medium that is used to generate the light.Atomic clocks that operate on optical transitions in trapped ions or atoms are the most accurate instruments ever made by mankind [1][2][3][4]. In an atomic clock, the quantum mechanical phase between two atomic levels oscillates at a frequency given by the energy difference between those levels. The atom's oscillation is detected by a laser, and today's best clocks are limited by the frequency stability of that readout laser. Recently Murray Holland and colleagues have proposed an ultrastable laser based on an atomic gain medium with a very narrow frequency response [5]. The paper by Bohnet et al. on p. xxx of this issue [6] now reports the first prototype and characterizes key features of such a system. Atomic transitions used for optical clocks have linewidths at the Millihertz level and standard freerunning lasers are not sufficiently stable in frequency to directly interrogate the ultranarrow atomic transition. Rather, the world's best clock lasers are stabilized to a meticulously crafted and controlled reference optical resonator. Currently such reference resonators achieve relative frequency stability below 10 -15 , corresponding to a change in the resonator length by less than the radius of a proton. At this level, the stability of the reference resonator is limited by a fundamental process -thermal noise in the mirrors that leads to fluctuations in the length of the resonator, and thus in the frequency of the laser locked to it. Further progress, though difficult, may be possible by cryogenic operation and use of mirror materials with improved mechanical properties.As an alternative, theorists at Boulder have proposed a laser operating in an unusual parameter regime where the linewidth of the atomic gain medium is much smaller than the linewidth of the laser resonator. In such a system, that may be termed superradiant laser following Dicke's early proposal for a mirrorless laser [7], the laser oscillation is stored predominantly inside the atoms themselves, rather than in the light field circulating inside the resonator. This makes the laser frequency largely immune to cavity length changes (see Fig. 1), with an isolation that is given by the ratio of atomic to resonator linewidths. For their system Bohnet et al. measure an immunity factor exceeding 10 4 .To realize narrow-frequency gain in a medium consisting of rubidium atoms, a species that is easy to laser cool but where no suitable transition is readily available, Bohnet et al. resort to a trick: A narrow atomic line can be mimicked by applying an external laser to weakly drive a transition between two long-lived atomic ground states via a detuned excited state, see Fig. 1c. In this case, the gain profile, and thus the emitted laser light, is not absolutely narrow in frequency, but only when measured relative to the driving lase...