Prolonged exposure of colorless dry sodalite to alkali vapor causes the material to gradually turn blue, dark blue, and finally black. The blue color observed at low sodium uptake appears because the absorbed sodium atoms are spontaneously ionized. The electron produced by ionization is shared by the four sodium ions present in the sodalite cage (three initially there and the fourth originating from the absorbed atom). The color center created in this way is represented by the formula (Na+)(4)eF(3-). Here, e stands for the electron and F3- for the negatively charged frame surrounding a zeolite cage. At the highest loading, when each cage contains an absorbed alkali atom, the color centers are arranged in a body-centered cubic lattice, allowing the electrons associated with the centers to form bands. This may explain the black color observed at high concentration. In this paper we present measurements of the absorption coefficient of the black sodalite for photon energies between 0 and 3 eV, and interpret them by performing one-electron band structure calculations for a fully loaded compound. These calculations deal only with the "solvated" electrons. The effect of the other electrons is taken into account through an empirical potential energy representing the interaction of a solvated electron with the zeolite frame. Because of this we study only the bands formed by the electrons of the color centers. Since the gap in the electron energy bands of the dry sodalite is over 6 eV, the color of the black sodalite is controlled by the solvated-electron bands formed in this gap. The measured spectrum has a threshold of about 0.6 eV which seems to suggest that the system has a gap in the electronic structure and is therefore a semiconductor. The calculations indicate, however, that, if the one-electron picture is valid, the fully doped black sodalite is a narrow-band metal. The threshold in the spectrum appears because the transition matrix element is zero for transitions responsible for photon absorption, and not because of a gap in the density of states. The calculated spectrum is in reasonable agreement with the measured one. Conclusions based on one-electron calculations can be altered by electron-electron interactions, which could turn a metal into an insulator. Two simple criteria, proposed by Mott and Hubbard, were used to test whether this transition might occur in our system. Unfortunately the results indicate that the system is close to the transition region which means that predictions made by these simple criteria are not reliable.