Normally samples are only single or multiple. But there is a specific situation in which an interpolated sample is used to correct a graph that only occurs with event trends.
The frequency of the data stored in an event trend can vary dramatically; for example, where several samples are within one display period, followed by no samples for a long time. A multiple sample will be drawn with a value calculated from the samples within the period. But the value after that period will be whatever the last sample in the period was. So an interpolated sample is added at the start of the next display period to correct the graph.