The basis for the anomalies of water is still mysterious. Quite generally tetrahedrally coordinated systems, also silicon, show similar thermodynamic behavior but lack-like water-a thorough explanation. Proposed models-controversially discussed-explain the anomalies as a remainder of a first-order phase transition between high and low density liquid phases, buried deeply in the "no man's land"-a part of the supercooled liquid region where rapid crystallization prohibits any experimental access. Other explanations doubt the existence of the phase transition and its first-order nature. Here, we provide experimental evidence for the first-orderphase transition in silicon. With ultrashort optical pulses of femtosecond duration we instantaneously heat the electronic system of silicon while the atomic structure as defined by the much heavier nuclear system remains initially unchanged. Only on a picosecond time scale the energy is transferred into the atomic lattice providing the energy to drive the phase transitions. With femtosecond X-ray pulses from FLASH, the free-electron laser at Hamburg, we follow the evolution of the valence electronic structure during this process. As the relevant phases are easily distinguishable in their electronic structure, we track how silicon melts into the lowdensity-liquid phase while a second phase transition into the high-density-liquid phase only occurs after the latent heat for the first-order phase transition has been transferred to the atomic structure. Proving the existence of the liquid-liquid phase transition in silicon, the hypothesized liquid-liquid scenario for water is strongly supported.liquid-liquid hypothesis | silicon phases | ultrafast spectroscopy | X-ray spectroscopy W ater exhibits more than sixty anomalous properties (1), and controversy about the exact physical origin-based in details of the atomic and electronic structure of water-has risen again some years ago (2-5). Although molecular dynamics simulations established good agreement with most experiments (6-8), new findings suggested that the theoretical models need to be refined to agree with improved experimental datasets (9-14). The recent results hinted to a more complex nature of water at standard conditions. In the following years, readjusted theoretical models have been presented (15)(16)(17). Most of the discrepancies could be solved, but still several concurrent models, partly contradicting each other, are compatible with the present data (18-21). Experiments able to decide among the models need to conquer the "no man's land" in water. This area of the phase diagram is inaccessible with equilibrium methods, because water rapidly crystallizes before the transient phases under question appear. The dispute could be solved by either establishing the existence of a critical point in this region, connected to a first-order phase transition between the low-density liquid (LDL) and high-density liquid (HDL) phases, or to rule out the existence of such a phase transition (22,23). Therefore, a lot of experimenta...