Case studies in scientific reproducibility