[NYSAA (1986)] Based on comparative study of the teeth, it seems that a minimum of 3, possibly 4 deer were butchered at the site. Their ages at death were: 1) 1 year, 4-5 months; 2) 2 1/2 years; 3) 8-10 years; 4) more than 6 years (possibly the same individual as #3). Deer generally give birth in late May or early June; thus, we can infer that deer #1 was killed in September or October, and deer #2 was probably killed around November. These death-dates support the assumption that rockshelters were occupied primarily during the fall and winter. Point typology indicates that the earliest occupation of the Ossining Rockshelter occurred c. 3400-2300 B.C. (uncalibrated dates for the Vosburg complex). The site was most frequently used by makers of Narrow Stemmed points (c. 2400-1600 B.C.). The catch-all category "Narrow Stemmed" used here subsumes several types that have been referred to by different names in the literature on Northeastern prehistory: Lamoka, Wading River, Sylvan Stemmed, Taconic, and Bare Island. These distinctive names may correspond, to some extent, to regionally varying cultural traditions. However, Funk (1976:158) found that the 173 points from Stratum 2 of the Sylvan Lake Rockshelter, which could be broken down into groups corresponding to named types, nevertheless belonged to a single component, and "had obviously been made and used by a single people" (1976:247). He suggests that the term "Sylvan Stemmed" should be applied to all such points from the Hudson Valley. Funk leaves open the question whether the observed morphological variation should be regarded as essentially random, or might reflect functional, chronological, or micro-traditional factors. Brennan (1967) had previously noted variation in shape and stem form in points he assigned to the "Taconic" tradition, and sought to arrange the variants as a chronological series. Funk (1976:248) notes that Kinsey, Holzinger, and Kent, who examined points from the Wading River, Hornblower, and Sylvan Lake sites, agreed that each assemblage included "one 42 THE BULLETIN or two" Bare Island points, whose average length is 2 inches. Four of the stemmed points from the Ossining Rockshelter can be classified, on the basis of size and shape, as Bare Island points (Figure 3, #s 24 and 27; Plate 2, #s 2 and 3). The points in Figure 4 were associated with a Lackawaxen Stemmed point (#1), in the cache (described in detail in Fiedel 1984). The Lackawaxen type seems to have been indigenous to the Delaware Valley and intrusive in the lower Hudson region. Similarly, Bare Island points appear to be more characteristic of Late Archaic assemblages in Pennsylvania and New Jersey than in New York. In my earlier paper dealing with the implications of the cache, I suggested that the Lackawaxen might have reached the Hudson Valley after a series of inter-band exchanges. The occasional presence of Bare Island points at sites in eastern New York may be ascribed to similar exchanges, although we cannot rule out the possibility that northward-migrating groups may have penetrated the Hudson Valley from time to time. The Ossining cache might conceivably represent such a group's brief occupation of the rockshelter. The chronological significance of the narrow stemmed points also calls for some discussion. Recently, narrow stemmed points have reportedly been found in association with Woodland ceramics at several sites in southern New England. If these associations bear up under further scrutiny, we will no longer be able to treat narrow stemmed points as reliable indicators of the Late Archaic age of sites that cannot be dated by other means. However, it should be noted that the great majority of stemmed points found by Funk (1976) at Sylvan' Lake and several other stratified sites were restricted to strata of Late Archaic age. The stemmed points from the Ossining Rockshelter display a degree of stylistic uniformity that suggests that they might have been produced over the coarse of a few hundred years perhaps, but not 3 millennia. I therefore continue to assign these points to the period from 2400 to 1600 B.C. A Normanskill-like point, and a Dry Brook or Susquehanna-like point, which resemble one another enough to be regarded as variants of a single type, indicate occasional occupation of the shelter around 1900-1500 B.C. (River phase). The Snook Kill base dates to about 1600-1500 B.C., and the complete Orient Fishtail point and basal fragments can be dated to about 1000-700 B.C. There are no points in the collection that can be assigned to the Early or Middle Woodland periods. One Levanna point indicates occupation during late Middle Woodland or Late Woodland times (c. A.D. 9001300); the potsherds found at the site also date to this period. The apparent gap in occupation from 700 B.G. to A.D. 900 is consistent with the general pattern observed at other rockshelters in the Hudson Valley (Funk 1976). Possible explanations of this pattern