|
Article Excerpt
ABSTRACT. The marine reservoir effect is known to skew radiocarbon dating (marine samples appear "older" than terrestrial samples of equivalent age), but the magnitude of this effect is not the same in all locations. Carbon-14 age determinations from 23 paired samples of terrestrial and marine origin are presented for five areas around the northern and eastern Bering Sea. It appears statistically suitable to average the age differences for three pairs, weighted inversely by variance. Differences from within each of 14 pairs of wood charcoal vs. sea mammal residue (from St. Lawrence Island, Cape Prince of Wales, Nunivak Island, Alaska Peninsula, and Unalaska Island) yield a weighted mean of 737 [+ or -] 20 years. Somewhat more variant differences from within each of five additional pairs of wood charcoal vs. sea mammal residue (St. Lawrence Island, Unalaska Island) provide a lower weighted mean of 460 [+ or -] 41 years. Differences from within each of four pairs of wood charcoal vs. marine shell (Nunivak Islan d) produce a weighted mean of 459 [+ or -] 32 years. Variations in these apparent reservoir effects presumably result largely from the interplay of differential ocean water upwelling and customary faunal feeding areas, although possible effects of other species characteristics cannot be ruled out. Dating of marine samples from the Bering Sea should thus either proceed with the expectation that age determinations may not be accurate within several centuries, or be approached through experimental measurement of reservoir effect among restricted faunal species in limited areas. Overall, the effect throughout the eastern Bering Sea appears to range from about 450 to 750 years.
Key words: Alaska, Bering Sea, reservoir effect, radiocarbon, differential ages, archaeology, shell, marine mammal
RESUME. On sait que l'effet du reservoir marin biaise la datation au radiocarbone (les echantillons marins semblent etre [much less than]plus vieux[much greater than] que les echantillons terrestres d'age similaire), mais la grandeur de cet effet n'est pas la meme a tous les endroits. On presente les determinations de l'age au carbone 14 pour 23 echantillons apparies d'origine terrestre et marine provenant de cinq zones situees aux environs de la mer de Bering septentrionale et orientale. Il semble approprie sur le plan statistique d'etablir la moyenne des differences d'age pour trois paires, ponderees inversement par la variance. Les differences provenant de chacune des 14 paires de charbon de bois compare a des restes de mammifere main (ile Saint-Laurent, cap Prince-de-Galles ile Nunivak, peninsule d'Alaska, et ile Unalaska) donnent une moyenne ponderee de 737 [+ or -] 20 ans. Des differences un peu plus variables provenant de chacune de cinq autres paires de charbon de bois compare a des restes de mammifer e main (ile Saint-Laurent, ile Unalaska) donnent une moyenne ponderee plus faible de 460 [+ or -] 41 ans. Les differences entre chacune des quatre paires de charbon de bois compare a des coquillages marins (ile Nunivak) donnent une moyenne ponderee de 459 [+ or -] 32 ans. Les variations dans ces effets de reservoir apparents proviennent probablement en grande partie du jeu reciproque entre les zones de remontee differentielle des eaux oceaniques et les aires d'alimentation habituelles, bien qu'on ne puisse eliminer les effets possibles de caracteristiques propres a d'autres especes. La datation des echantillons marins de la mer de Bering devrait donc se faire soit en s'attendant a des erreurs de plusieurs siecles dans les determinations d'age, soit en l'abordant par le biais de mesures experimentales portant sur l'effet de reservoir parmi un petit nombre d'especes fauniques et a l'interieur d'un territoire restreint. Dans l'ensemble, cet effet dans tout l'est de la mer de Bering semble aller de 450 a 750 ans.
Mots cles; Alaska, mer de Bering, effet de reservoir, radiocarbone, ages differentiels, archeologie, coquillage, mammifere main
Traduit pour la revue Arctic par Nesida Loyer.
INTRODUCTION
It is now well known that radiocarbon dating of marine samples--shell or marine mammal residue--is skewed by the reservoir effect of the oceans, so that in most regions marine samples yield radiocarbon ages substantially older than those yielded by terrestrial samples that are in fact equivalent in true calendar age. Although data regarding magnitude and regional variations of this effect have been presented for a number of regions of the world, no presentation has specifically targeted the Bering Sea. Here, we present suites of archaeologically derived dates on paired samples of terrestrial and marine origin from five areas in and around the northern and eastern Bering Sea. The St. Lawrence Island dates are based on samples collected nearly three-quarters of a century ago but dated recently, and we also consider some ages obtained there in the 1970s. Excavation dates of samples from four other locations were as follows: Nunivak Island, 1996 and 1997; lower Naknek River on Bristol Bay, 1998; Cape Prince of Wa les, 1998 and 1999; Umnak Island of the eastern Aleutians, 1998.
Archaeologists working in the North first became aware of the problem of determining site age on the basis of marine products some 25 years ago, when it was pointed out that ages from seal products systematically "predated" those obtained from wood charcoal samples in apparently contemporary contexts (e.g., McGhee and Tuck, 1976). Since that time, a number of studies have presented information bearing on the question, with some particular attention paid to the dating of marine shells collected on dates known historically (e.g., Robinson and Thompson, 1981). We are aware of only one previous attempt with shellfish from the Bering Sea, involving shell of Astarte borealis dredged live in 1969 from a spot 50 km or so south-southwest of Cape Prince of Wales, which in 1974 yielded the relatively imprecise (and unpublished) radiocarbon age of 540 [+ or -] 200 years (sample W-2768; R. Rowland, pers. comm. 1999).
Stated briefly, atmospheric (14) C results from cosmic bombardment, the magnitude of which varies in response to short-term fluctuations in the strength of magnetic fields, both terrestrial and interplanetary. Although the worldwide level of radioactive carbon that results and enters into atmospheric carbon dioxide is roughly constant at a given time, the speed and degree of exchange with the ocean are influenced by water depth, so that radiocarbon measurements tend to vary inversely with distance from surface. Materials from surface waters--the upper 75--199 in--measure on worldwide average about 400 years too "old" compared to terrestrial samples, whereas those from greater depths measure correspondingly "older" (Oeschger et al., 1975; Stuiver and Braziunas, 1993). At the same time, however, the effect in the various levels of the ocean is to dampen reflections of short-term oscillations that occur in the incidence of atmospheric (14) C (see, for instance, Stuiver and Braziunas, 1993: Fig. 14).
This phenomenon constitutes the oceanic reservoir effect. But it has been clearly shown that substantial regional variation in the magnitude of this effect in surface waters results from the degree of local upwelling, which brings deeper waters into the upper levels (Taylor, 1987:126-- 131; Bowman, 1990:24--25; Stuiver and Braziunas, 1993). In this connection, all waters entering the Bering Sea proceed from the Alaska Current that moves counterclockwise around the Gulf of Alaska, flowing southwestward over (and possibly through) the deep Aleutian Trench, from which some waters turn north through various channels between the Aleutian Islands. After circling counterclockwise in the Bering Sea, especially southwest of the edge of the Bering Platform, where some depths exceed 3000 m, a portion of these waters exits between the westernmost Aleutians and the Kamchatka Peninsula, whereas other portions flow more steadily northward above the Bering Platform and through Bering Strait into the Chukchi Sea (U.S. Depart ment of Commerce, 1961).
One of the major discussions of the reservoir problem, oriented toward achieving a means for reconciling marine-based determinations not only with those from terrestrial samples, but also with the modern calendar (Stuiver and Braziunas, 1993), recognizes the existence of worldwide variations in the reservoir effect. However, it proposes an overall model of the effect in the surface layers of the ocean that mitigates the impact of short-term fluctuations in atmospheric (14) C by attempting to parallel the damping effect of the ocean on such fluctuations in the atmosphere. A computer program (CALIB) incorporates the model into a calendrical calibration routine (Stuiver and Reimer, 1993). Using the model, in which the overall effect is to reduce (14) C ages on marine samples by a...
|
