released oligomeric phosphoribosyl-AMP into ethanol-soluble form. The ethanol was removed by evaporation under reduced pressure. The resulting aqueous solution was buffered at pH7.4 with Tris, and treated with snake-venom phosphodiesterase to give 32Pand 3H-labelled AMP and phosphoribosyl-AMP, which were both shown to be present by the methods described above. Experiments with [3H]adenosine instead of [32P]Pi showed similar results. In this case the isolated nuclei were not incubated with [3H]NAD+. The chain-lengths of protein-bound polymer from whole nuclei, buffer-soluble and CaC12-soluble fractions in vitro were estimated by using Dowex 1 (Clform) chromatography (Fujimura & Sugimura, 1971). The average chain lengths were about 5-7 units in all three cases (O’Farrell, 1973). We believe that the chain length of the poly(ADPribose) synthesized in vivo in P. polycephalum is between 2 and 4 units. It is possible that the ADP-ribosylation of nuclear protein in vitro may not be a true reflection of the situation in vivo; perhaps in intact cells fewer, or only one, protein serves as an acceptor.