More than 30% of oceanic crust is formed at fast-spreading ridges, but there remains much to learn about how the lower, gabbroic crust is accreted and cooled. This is partially due to conflicting evidence from direct and indirect studies of the lower crust and ophiolite analogues, leading to two end-member models for gabbroic crust formation which have different predictions for the thermal structure of the crust. The gabbro glacier model asserts the majority of heat is lost through the upper crust via hydrothermal circulation while the gabbroic crust cools conductively off-axis. The sheeted sills model requires efficient cooling throughout the crust within 2-3 km of the spreading center via deep hydrothermal circulation. These models are testable using magnetic techniques. As the lower ocean crust cools, magnetic minerals will lock in the magnetic field direction. If the crust is cooling during a magnetic reversal, the boundary between polarity intervals acts as a proxy isotherm in the lower crust. The slope of that isotherm will distinguish between the gabbro glacier model, which should have shallow isotherms dipping away from the ridge, and the sheeted sills model, where isotherms would steeply dip near the ridge. A 2017 cruise to Pito Deep, a tectonic exposure in fast-spread crust, recorded near-bottom magnetic data and collected oriented gabbroic samples to document a polarity boundary in the lower crust. This thesis developed a magnetic inversion technique to incorporate near-bottom anomaly data from multiple platforms over steep escarpments to identify the location of the polarity boundary at depth. Results show a large offset of the polarity boundary between the dike and gabbroic layers, implying that the gabbroic layer remains hotter for an extended period off-axis. Using oriented samples from a 1x1 km study area, I verified the location of a polarity boundary which remains horizontal for a minimum of 8 km off-axis. Finally, I use samples with multiple polarity components to determine a cooling rate in gabbroic crust. Our results are incompatible with deep hydrothermal cooling within 2-3 km of the spreading center, and instead promote the idea of slow cooling within 8 km of the spreading center.