Surface and deepwater paleoclimate records in Irminger Sea core SO82‐5 (59°N, 31°W) and Icelandic Sea core PS2644 (68°N, 22°W) exhibit large fluctuations in thermohaline circulation (THC) from 60 to 18 calendar kyr B.P., with a dominant periodicity of 1460 years from 46 to 22 calendar kyr B.P., matching the Dansgaard‐Oeschger (D‐O) cycles in the Greenland Ice Sheet Project 2 (GISP2) temperature record [Grootes and Stuiver, 1997]. During interstadials, summer sea surface temperatures (SST_su) in the Irminger Sea averaged to 8°C, and sea surface salinities (SSS) averaged to ∼36.5, recording a strong Irminger Current and Atlantic THC. During stadials, SST_su dropped to 2°–4°C, in phase with SSS drops by ∼1–2. They reveal major meltwater injections along with the East Greenland Current, which turned off the North Atlantic deepwater convection and hence the heat advection to the north, in harmony with various ocean circulation and ice models. On the basis of the IRD composition, icebergs came from Iceland, east Greenland, and perhaps Svalbard and other northern ice sheets. However, the southward drifting icebergs were initially jammed in the Denmark Strait, reaching the Irminger Sea only with a lag of 155–195 years. We also conclude that the abrupt stadial terminations, the D‐O warming events, were tied to iceberg melt via abundant seasonal sea ice and brine water formation in the meltwater‐covered northwestern North Atlantic. In the 1/1460‐year frequency band, benthic δ¹⁸O brine water spikes led the temperature maxima above Greenland and in the Irminger Sea by as little as 95 years. Thus abundant brine formation, which was induced by seasonal freezing of large parts of the northwestern Atlantic, may have finally entrained a current of warm surface water from the subtropics and thereby triggered the sudden reactivation of the THC. In summary, the internal dynamics of the east Greenland ice sheet may have formed the ultimate pacemaker of D‐O cycles.

We present two new high-resolution sediment records from the southwestern Iceland and Norwegian Seas that were dated by numerous ¹⁴C ages up to 54 ¹⁴C ka bp. Based on various lines of evidence, the local ¹⁴C reservoir effect was restricted to 400–1600 yr. The planktic stable isotope records reveal several meltwater spikes that were sampled with an average time resolution of 50 yr in PS2644 and 130 yr in core 23071 during isotope stage 3. Most of the δ¹⁸O spikes correlate peak-by-peak to the stadials and cold rebounds of the Dansgaard-Oeschger cycles in the annual-layer counted GISP2 ice core, with the major spikes reflecting the Heinrich events 1–6. This correlation indicates large fluctuations in the calibration of ¹⁴C ages between 20 and 54 ¹⁴C ka bp. Generally the results confirm the ¹⁴C age shifts as predicted by the geomagnetic model of Laj, Mazaud and Duplessy (1996). However, the amplitude and speed of the abrupt decrease and subsequent major increase of our ¹⁴C shifts after 45 ¹⁴C ka bp clearly exceed the geomagnetic prediction near 40–43 and 32–34 calendar (cal) ka bp. At these times, the geomagnetic field intensity minima linked to the Laschamp and the Mono Lake excursions and confirmed by a local geomagnetic record, probably led to a sudden increase in cosmogenic ¹⁴C and ¹⁰Be production, giving rise to excess ¹⁴C in the atmosphere of up to 1200%.