Sediment cores were recovered during the ship expedition of German RV Polarstern in 2016 (PS97) using piston corers. For paleo- and rock magnetic analyses clear plastic boxes of 20×20×15 mm were pressed into the split halves of the generally 1 m long sections of the sediment cores.

In order to determine the direction of the characteristic remanent magnetization (ChRM), demagnetization results of the NRM were subjected to principal component analysis (PCA) according to Kirschvink (1980). The PCA also provided the maximum angular deviation (MAD) as a measure of the precision of the determined ChRM direction. ChRM declinations obtained by PCA were rotated around a vertical axis until the declinations of all samples falling into a circular window of 35° around the direction expected from a geocentric axial dipole (-72.9°) yielded a mean of 0°. ChRM data from core PS97-085-1 (-85-3) were tentatively tilted by +17° (-7°) around the EW axis in order to parallel the maximum in the inclination distribution with the inclination of a geocentric axial dipole field.

The anhysteretic susceptibility K(ARM) is defined as the ARM intensity normalised by the static field used for producing the ARM. The anhysteretic susceptibility normalised by the low field bulk susceptibility K(ARM)/klf then is a magnetic grain size proxy with low (high) ratios indicating relatively large (small) magnetite particles.

In order to discriminate samples being dominated by low-coercive minerals (magnetite, Fe3O4 and greigite, Fe3S4) from samples being dominated by high-coercive minerals (mostly hematite, Fe2O3), the S-ratio was calculated using S=0.5×(1-[IRM(-200 mT)/SIRM(1500 mT)]). S-ratios range from 0 to 1, with: dominance of magnetite/greigite: 0<<S≤1, and dominance of hematite: 0≤S<<1.

As another grain size proxy the ARM intensity was normalised by the SIRM: (1000×ARM/SIRM) with low (high) ratios indicating relatively large (small) magnetite particles. The factor of 1000 is introduced in order to avoid small numbers.

Relative paleointensity variations were estimated by three different proxies: slope of NRM vs. ARM of common demagnetization steps (slope(NRM/ARM)), NRM intensity demagnetized with 30 mT normalized with bulk susceptibility klf (pjk(30mT)), and NRM intensity demagnetized with 30 mT normalized with saturation magnetization SIRM (pjs(30mT)).

Data records were turned into time series by correlation to dated reference records from Antarctica (Wu et al., 2021) and the Black Sea (Liu et al., 2021).