We analyse the results of Thellier palaeointensity experiments modified so that the measured magnetisation at any stage may be decomposed into components of natural remanent magnetisation (NRM) and laboratory thermal remanent magnetisation (TRM). We demonstrate that the very long high temperature tails of pTRM acquired by multidomain (MD) grains may be detected in Thellier experiments without recourse to explicit pTRM tail check measurements. This can save time in the experimental process and provides a more sensitive indication of how MD behaviour may affect that particular experiment than either pTRM tail checks or the observation of zigzagging in an IZZI experiment. We observe that the action of imparting a pTRM tail also involves some removal of NRM. However, the pTRM tails are not the analogues of classical overprints as may be intuitively expected but instead appear to favour remagnetisation over demagnetisation. This and other observed behaviour is not consistent with any phenomenological model of MD TRM but is consistent with predictions made by kinematic theory.