Earth’s climate is intricately linked to the carbon cycle through the radiative effect of atmospheric carbon dioxide. The ocean plays a central role in this climate-carbon system; as oceans store ∼50 times more carbon than the atmosphere, even small changes in ocean chemistry could greatly affect global climate. Understanding how the oceanic carbon reservoir has evolved across changing climates is thus critical for both constraining mechanisms of climate change and predicting impacts from anthropogenic carbon addition. This dissertation contributes to knowledge of the ocean carbon reservoir’s evolution across the last 1.5 million years of Earth’s history, with a particular focus on two key intervals of climatic change: 1) Present day, when a large, human-sourced perturbation to the carbon cycle is underway, the effects of which are not yet fully realized; and 2) The mid-Pleistocene transition (MPT; ∼900,000 years ago), when natural cycles of global warming and cooling increased in intensity and duration. Without direct observations for both these time intervals, I focus on documenting changes to ocean carbon chemistry using proxies for seawater composition. The primary tools for this purpose are boron concentrations (B/Ca ratios) and the boron isotopic composition (δ11B) of carbonate skeletons produced by marine organisms. These tools are rooted in the aqueous chemistry of boron, in which the speciation and isotopic composition of boron compounds change with seawater pH. To test present-day changes in the oceanic carbon reservoir, I measured δ11B on the calcitic skeletons of deep-sea corals (genus Keratoisis). Results show that while coral δ11B does correlate with deep ocean pH, δ11B variations within coral skeletons are too large to be explained by changes in deep ocean pH over the corals’ lifespan. These variations most likely reflect the biology of the coral organism, suggesting that δ11B measurements in Keraotisis cannot be utilized to track ocean pH until coral growth mechanisms are better understood. To complement these δ11B data, I measured the radiocarbon (14C) content of Keratoisis skeletons. Results show that coral skeletal 14C tightly correlates to the 14C content of the deep ocean, and that bamboo corals live for 50 to 300 years with radial growth rates of 10 to 80 μm per year. This supports the use of 14C for generating bamboo coral ages and growth rates, and for tracking perturbations to the 14C content of the deep ocean. Through my deep-sea coral study, I learned the importance of accurate and precise δ11B measurements for sound interpretations of ocean carbon chemistry. These interpretations necessitate highly specialized analysis protocols. While two protocols are commonly applied for δ11B measurements, existing comparisons found relatively large offsets between both protocols. To trace the cause and implications of this offset, I established a new δ11B measurement protocol and performed an internal comparison between the new and existing measurement protocols. Results confirm that carbonate δ11B values are significantly offset between techniques. Although the nature of this offset remains enigmatic, I show that both techniques show the same δ11B-to-pH sensitivity, and consistent pH estimates are obtained when a protocol-specific constant offset is applied. This suggests that both δ11B analysis protocols can be applied for reconstructing pH with equal confidence. To test for changes in the ocean carbon reservoir across the MPT, I investigated the B/Ca and Cd/Ca composition of the benthic foraminifer Cibicidoides wuellerstorfi to track deep ocean carbonate saturation state (∆[CO32−]) and nutrient inventories. At 4.3 km water depth in the South Atlantic Ocean, B/Ca abruptly decreased by 20% and Cd/Ca increased by 40% between 950 and 900 ka, equivalent to a 60 μmol/kg increase in abyssal ocean carbon storage. Coincident shifts in deep ocean circulation and atmospheric pCO2 around 900 ka suggest that a new regime of deep ocean carbon sequestration developed during the MPT. I argue that this regime was intricately linked with the increased magnitude and duration of glacial cycles following the MPT.