Newly derived fossil pollen data were obtained from four sites along a transect from the boreal forest limit into tundra in the eastern Northwest Territories. Multivariate statistical analyses were employed to interpret the pollen assemblages. Transfer functions were constructed between the pollen data and climatic data, and paleoclimatic estimates were derived. The objective nature of the reconstructions provides an independent verification of the general outlines of the chronology of tree-line movements during the mid- and late-Holocene as established in previous paleosol and pollen studies. Boreal forest extended to approximately 62°N, associated with mean July temperatures 1 to 3°C above modern means, from at least 5500 to 3700 yr B.P. Although a major episode of southward displacement of tree line at about 3700 yr B.P. is recorded, later events are not clearly represented. Considerations of the statistics, the time scales, and the nature of the pollen rain suggest only conservative interpretations of the results are possible. It is suggested that the pollen sites may have been sensitive recorders of regional vegetation change only when they were near the ecotone, corresponding to a climatic threshold.

The distribution of quartz in the surface sediments of the Atlantic Ocean reflects derivation from continents by means of rivers, wind, ice, and coastal erosion. Enrichment of quartz thus supplied has occurred in some deep basins of especially the southern high latitudes from winnowing of finegrained clays by bottom currents. Although similar modes of quartz transport may have operated both during the Holocene and the last glacial maximum (18,000 yr B.P.), significant differences in the intensity of transport and in the locii of deposition, which are attributable to climatic variations during these times, exist in some areas of the Atlantic. In Holocene sediments of the eastern equatorial Atlantic, a band of high percent quartz exists directly off the present Saharan Desert and Sahel region and reflects the trade-wind transport of dusts from these arid and semiarid regions. During the last glacial maximum (18,000 yr B.P.), this high quartz band expanded southward by about 8° of latitude. This expansion was caused not only by the southward expansion of aridity and desert dunes but also by the southward migration of the northerly belt of trade winds during the last glaciation. Relatively high abundances and accumulation rates of quartz during the last glaciation suggest higher intensities of trade winds during that time compared to the Holocene. In the North Atlantic, the abundances of quartz in Holocene sediments are high adjacent to Greenland-Iceland and in the areas off Newfoundland-Labrador, and gradually decrease toward the central areas. The polar front and limit of sea-ice melting are at present confined to the northern part of the North Atlantic. The ice-rafting of quartz grains is, therefore, effective in the areas adjacent to Greenland and to some extent off Labrador causing high abundances in these areas. In contrast to this, during the last glaciation, the quartz abundances and accumulation rates are high in the central areas of North Atlantic around 45°N and decrease toward Greenland-Newfoundland. The migration of the polar front to as far south as 45°N and the consequent southward migration of sea-ice melting and ice-rafting during the last glaciation apparently caused this change in distribution. In addition to ice-rafting at present, wave or current reworking of relict glacial-marine detritus may have caused the high abundance of quartz in the surface sediments off Newfoundland-Labrador. In 18,000 yr B.P. sediments of the Norwegian Sea, the area of high percent quartz (>10%) is more extensive than that in Holocene sediments. This reflects the greater influence of ice-rafting or glacier activity in the sediment dispersal in the Norwegian Sea during the last glacial times.

A series of closely spaced radiocarbon measurements on a carbonate-rich box core from the western equatorial Pacific show a mixed layer at least 7 cm thick, with 14C ages between 4000 and 5000 years, and an orderly progression of ages below this layer, indicating an average sedimentation rate of about 2 cm/103 yr. The profile can be simulated using a numerical extension of the mixing model of Guinasso and Schink (1975)and a numerical exponential mixing model. The best-fit iteration indicates an apparent mixing coefficient of K = 120 cm2/103 yr which also fits well the excess 210Pb distribution. The best-fit also indicates that a small amount of sediment was lost on the top, and that there was a reduction in sedimentation rate within the early Holocene.