Longitudinal studies are almost always plagued by missing data. Examples include research data in public health, medicine, life and social sciences, as well as in environmental and geophysical studies of ice ages and the past climate. In the case of data involving people over a period of time there are usually some drop-outs due to e.g. moving away, or death, or simply disinclination to continue. Similarly, some physical measurements in a historic time series may be corrupted or missing. In such cases it may be impossible to repeat the measurements, or the survey, so that the best that can be done is to find a way of accommodating the lacunæ such that they do minimal damage to the reliability of any conclusions.