Replenished axial melt lenses at fast-spreading mid-oceanic ridges may move upward and intrude into the overlying hydrothermally altered sheeted dikes, resulting in high-grade contact metamorphism with the potential to trigger anatexis in the roof rocks. Assumed products of this process are anatectic melts of felsic composition and granoblastic, two-pyroxene hornfels, representing the residue after partial melting. Integrated Ocean Drilling Program Expeditions 309, 312, and 335 at Site 1256 (eastern equatorial Pacific) sampled such a fossilized oceanic magma chamber. In this study, we simulated magma chamber roof rock anatectic processes by performing partial melting experiments using six different protoliths from the Site 1256 sheeted dike complex, spanning a lithological range from poorly to strongly altered basalts to partially or fully recrystallized granoblastic hornfels. Results show that extensively altered starting material lacking primary magmatic minerals cannot reproduce the chemistry of natural felsic rocks recovered in ridge environments, especially elements sensitive to hydrothermal alteration (e.g., K, Cl). Natural geochemical trends are reproduced through partial melting of moderately altered basalts from the lower sheeted dikes. Two-pyroxene hornfels, the assumed residue, were reproduced only at low melting degrees (<20 vol%). The overall amphibole absence in the experiments confirms the natural observation that amphibole is not produced during peak metamorphism. Comparing experimental products with the natural equivalents reveals that water activity (aH2O) was significantly reduced during anatectic processes, mainly based on lower melt aluminum oxide and lower plagioclase anorthite content at lower aH2O. High silica melt at the expected temperature (1000–1050 °C; peak thermal overprint of two-pyroxene hornfels) could only be reproduced in the experimental series performed at aH2O = 0.1.