Abstract-We report on a new angrite, Northwest Africa (NWA) 1296, a fine-grained rock with a magmatic texture of rapid cooling. Dendritic olivine (≈Fo 50 ) crystallized first in association with anorthite microcrysts (An 98-100 ) forming composite chains separated from one another by intergrown Al-Fe diopside-hedenbergite pyroxenes. In addition, some olivines with lower Mg# and increased CaO (up to 12%) are found between the chains as equant microphenocrysts. Pyroxenes and olivines are both normally zoned from Mg# = 0.52 to less than 0.01 in the rims. Ca-rich olivines are surrounded by, intergrown with, or replaced by subcalcic kirschsteinite. They appear after plagioclase crystallization stopped, at the end of the crystallization sequence. Minor phases are pyrrhotite, Fapatite, and titanomagnetite. Pyroxene is the last silicate phase to grow, interstitial to idiomorphic olivine-kirschsteinite. Numerous small vesicles and some channels are filled with microcristalline carbonate. The mode (vol%) is about 28% olivine, 3% kirschsteinite, 32% anorthite, 34% pyroxene, and 3% of the minor phases-close to that reported previously for D'Orbigny and Sahara (SAH) 99555. The bulk chemical composition of NWA 1296 is similar to D'Orbigny and SAH 99555; NWA 1296 differs by its texture and mineralogy, which are interpreted as resulting from rapid crystallization-an evidence of impact melting. Angrites cannot be produced by partial melting of a CV source because segregation of a "planetary" core is necessary to explain the low FeO/MgO ratio of magnesian olivines. Neither the odd Ca/Al ratio nor the very low SiO 2 content can be explained by conventional partial melting scenarios. We suggest that carbonate is the key to angrite genesis. This is supported by the striking similarities with terrestrial melilitites (low SiO 2 , superchondritic Ca/Al ratio, presence of carbonate). The lack of alkalies could be the result of either loss after impact melting or absence of alkalies in the source.