For quantum computing to become viable, the inherently fallible nature of qubits must be overcome with quantum error correction (QEC). QEC requires coherent qubits, in a configuration compatible with a given QEC scheme, and quantum logic operations with sufficiently low error. In this thesis, we perform a series of targeted experiments to achieve these goals on a path toward realizing the surface code QEC scheme. We first develop the Xmon variant of the transmon qubit, a highly coherent, planar, and frequency tunable superconducting qubit. With coherence demonstrated, we build an array of five Xmon qubits in a configuration compatible with the surface code, and demonstrate quantum logic operations with sufficiently low error to employ the surface code. These logic gates are characterized with randomized benchmarking, a protocol for determining gate error. We find applications of randomized benchmarking beyond the intended use in gate optimization and decoherence characterization, in addition to exploring the fundamental assumptions of randomized benchmarking. Lastly, we build a nine qubit Xmon transmon array and demonstrate correction of environmental bit-flip errors in a precursor to the surface code.