THE principal expense and source of difficulty in equipment for nuclear magnetic resonance is the magnet to provide the large magnetic field. It has been shown that a nuclear-resonance signal can be obtained using only the Earth's field1,2. This is the so-called free-precession experiment, in which the nuclear system is first polarized at right angles to the Earth's field by means of a simple current-carrying coil. After removal of this field, the nuclear magnetization precesses about the Earth's field (approximately 0.48 gauss) and induces an alternating voltage in the same coil at a frequency of about 2 kc./s., which may be observed after suitable amplification. The polarizing field serves the double purpose of producing a nuclear magnetization appropriate to a large field and hence a stronger signal to be observed in the smaller field, and also produces this polarization at right angles to the small field. The signal is attenuated by loss of phase-coherence of the precessing nuclei, due both to interactions between them with a time constant T 2, and to inhomogeneity in the Earth's field (T 2 *). The method can be used for measuring the decay time, T 2, which is a property of the material, by observing the decay of the precession signal, provided the inhomogeneity of the Earth's field is sufficiently small for T 2 * ≫ T 2. In many cases, for example, in liquids such as water containing mobile protons, T 2 may be several seconds and the corresponding field mhomogeneity must be less than 10 microgauss/cm. in order to measure T 2. This implies that the experiment cannot be performed in the laboratory and usually requires the sample coil to be placed out of doors away from all buildings (we have found a brick wall to be unacceptably magnetic in this sense).