Abstract. Base composition measurements were used as a criterion of RNA changes in goldfish brain. RNA synthesized during the acquisition of new swimming skills was found to have a uridine/cytidine ratio 20-80 per cent higher than that of RNA formed under nonlearning conditions. A variety of behavioral situations were used to demonstrate that these RNA changes were not due to such factors as intense physical exertion, passive responses to stress, or the intense electrical activity of the brain and convulsive behavior produced by KCl injections. The RNA changes were produced by two behavioral situations;(1) during the process of acquiring new swimming skills, and (2) as a result of attempts to master an impossible task. The results suggest that the modified RNA synthesis taking place during the acquisition of new behavioral patterns is probably not specific as to the particular information content being stored, but may be required for the consolidation step of new information storage.Introduction. The attempts to correlate brain RNA synthesis with the acquisition of new behavior patterns have led to two types of observations: (1) a general increase in RNA synthesis,'-3 and (2) the formation of new RNA species with a changed base composition.47 In a previous paper,' it was suggested that the synthesis of an RNA with an altered base composition was characteristic of the information-gathering state of goldfish brain. This type of RNA appeared to be required for the consolidation of learned behavior but not for the repair of effects due to stress or to a generalized activity of the animals. When protein synthesis was inhibited by puromycin, the goldfish could acquire new swimming skills just as rapidly as in the absence of drug but could not remember the task 22 hours later. Under these conditions no decrease in brain RNA synthesis was observed and no alteration in base composition occurred. From these observations, a changed base composition was thought to be an index of the state of brain function during the acquisition phase of learning. It was, however, noted that the use of drugs to dissociate stress and work effects from those due to learning is subject to a number of difficulties, including observations that hippocampal seizures are produced by puromycin8 and the possible occurrence of anoxia arising from the reported decrease in respiration of nervous tissue in the presence of puromycin and cycloheximide.9 In the present study, the above features of the behavioral experiment were examined in detail and nonpharmacological methods were devised to dissociate the information-gather-