Snmmary.-This study investigates the learning of continuous serial lists of nonsense syllables in terms of the strategies by which Ss select the syllables to be learned first. Three hypotheses were tested and verified by the results: ( a ) Ss select the first temporal item of a series as cognitively first when instructional and structural bases for differentiation are not provided by E; (b) Ss also tend to cognize a structurally isolated item as first in the series; and ( c ) the relative effectiveness of these two anchoring tendencies depends upon the position of the structurally isolated syllable within the temporal series. Results support the theory that the cognition of one or more "beginning" items establishes the serialposition curve because the anchors provided by these items, which are learned first, facilirate the acquisition of the remaining items.The serial-position curve of learning has been explained in terms of S's strategy in making a cognitive selection of the items to be learned first (e.g., Harcum, 1967; Jensen, 1962a). Harcum and Coppage (1965a) proposed that the serial-acquisition task progresses through two stages, the first of which is the differentiation of items, including the cognition of one item as the beginnkg of the series. The second stage occurs when the first-learned items become cn anchor for the association of other items. Thus, with group data the characteristic bowed curve is produced because Ss are consistent in choosing a given item for the cognitive anchor. Harcum and Coppage (1965a) attributed the skew of the curve also to Ss' tendency to make forward associations to this anchor more rapidly than they form backward associations.The selection of cognitive anchors may be based on temporal primacy, structural inhomogeneiry, and varioiis other attributes of items which direct S"s acten--tion to them. E can manipulate these factors, respectively, by varying relative starting position within the series, by structurally isolating an rrem, or by instructions. These bases of strategies are not mutually exclusive, of course; moreover, the choice of cognitive anchors may be idiosyncratic for the different Ss (Lippman & Denny, 1964).With the usual method of presentation, in which the intertrial interval (ITI) is greater than the interstimulus rnrerval (ISI), most Ss select the first and last items of the series as anchors, presumably because of the increased distinctiveness of these items due co proximity co the easily discriminable IT1