The simulated ocean carbon sink in each year in Earth System Models depends on the starting year of the simulation and the atmospheric partial pressure at that year. Simulations of the last decades and the next century, such as in the Global Carbon Budget or the Coupled Model Intercomparison Project Phase 5 (CMIP5) and 6 (CMIP6) often start in the middle of the 19th century to save computational time, although atmospheric CO2 already started to increase due to anthropogenic activity since 1765. If a simulation starts in 1850, for example, the spin-up simulation is performed with the higher atmospheric partial pressure in 1850. In these simulations, old water masses in the ocean contain more carbon compared to a simulations that would have started in 1765. The difference in the carbon content of deep ocean water masses affects the amount of carbon that is taken up throughout the entire simulation period when these water masses continously come back to the ocean surface, resulting in a too small simulated ocean carbon sink in simulations that start in 1850. However, the magnitude of this difference remains uncertain. Here, I made two sets of simulations with the Earth System Model Bern3D-LPX, one set starts in 1765 and the other one starts in 1850 to quantify this difference. Each set contains 9 different configurations of Bern3D-LPX with different ocean mixing rates and climate sensitivites to cover the wide range of ESMs in CMIP. The model forcing setup is described in detail in Terhaar et al. (2023) (Environ. Res. Lett. 18 (2023) 024033, https://doi.org/10.1088/1748-9326/acaf91). Only output until 2020 should be used.