Water anomalies still defy explanation. In the supercooled liquid, many quantities, for example heat capacity and isothermal compressibility κ T , show a large increase. The question arises if these quantities diverge, or if they go through a maximum. The answer is key to our understanding of water anomalies. However, it has remained elusive in experiments because crystallization always occurred before any extremum is reached. Here we report measurements of the sound velocity of water in a scarcely explored region of the phase diagram, where water is both supercooled and at negative pressure. We find several anomalies: maxima in the adiabatic compressibility and nonmonotonic density dependence of the sound velocity, in contrast with a standard extrapolation of the equation of state. This is reminiscent of the behavior of supercritical fluids. To support this interpretation, we have performed simulations with the 2005 revision of the transferable interaction potential with four points. Simulations and experiments are in near-quantitative agreement, suggesting the existence of a line of maxima in κ T (LMκ T ). This LMκ T could either be the thermodynamic consequence of the line of density maxima of water [Sastry S, Debenedetti PG, Sciortino F, Stanley HE (1996) Phys Rev E 53:6144-6154], or emanate from a critical point terminating a liquid-liquid transition [Sciortino F, Poole PH, Essmann U, Stanley HE (1997) Phys Rev E 55:727-737]. At positive pressure, the LMκ T has escaped observation because it lies in the "no man's land" beyond the homogeneous crystallization line. We propose that the LMκ T emerges from the no man's land at negative pressure.scenarios for water | Widom line | Berthelot tube W ater differs in many ways from standard liquids: ice floats on water, and, upon cooling below 48C, the liquid density decreases. In the supercooled liquid, many quantities, for example heat capacity and isothermal compressibility, show a large increase. Extrapolation of experimental data suggested a powerlaw divergence of these quantities at −458C (1). Thirty years ago, the stability-limit conjecture proposed that an instability of the liquid would cause the divergence (2) (Fig. 1A). This is supported by equations of state (EoSs), such as the International Association for the Properties of Water and Steam (IAPWS) EoS (3), fitted on the stable liquid and extrapolated to the metastable regions. Ten years later, the second critical point interpretation, based on simulations (4), proposed that, instead of diverging, the anomalous quantities would reach a peak, near a Widom line (5, 6) that emanates from a liquid-liquid critical point (LLCP) terminating a first-order liquid-liquid transition (LLT) between two distinct liquid phases at low temperature (Fig. 1B). The two scenarios differ in the shape of the line of density maxima (LDM) of water ( Fig. 1 A and B). A recent work (7) has added one point on this line at large negative pressure, but this was not enough to decide between the two scenarios.It has been argued (8) that ...