It is possibly anti-climatic that the entire topic of this website can be answered on one page, but the evidence showing that plumes have drifted over time with respect to the spin axis is fairly conclusive. This though may mean that the spin axis may have drifted rapidly in responce to some other physical event (for example in True Polar Wander caused by centrifugal redistribution of mass, or alternatively the spin axis has remained constant and plumes have drifted in responce to some other mechanism (e.g. Mantle Wind)
The initial observations of plume drift came from the Hawaii-Emperor chain, but it has subsequently been observed in other Pacific island chains (see Sager and Koppers case study)
Hawaii -example of palaeolatitude shift
In 1997 John Tarduno (Tarduno et al. 1997)presented some startling finds from ODP leg 145 (1995) which showed that global plate motion reconstructions do not agree with a fixed hotspot being the source for the Hawaii/Emperor seamount chain. Instead his results suggested rapid motion of the Hawaiian hopspot at rates of up to 50mm/yr for short periods of time. The evidence for this came from palaeolatitude studies which should have showed the same latitude of formation for all the islands if the hotspot is fixed over time (i.e. fixed with respect to the poles)
In 2003 Tarduno published an updated study based on ODP leg 197 which was largely inspired by the earlier observations and aimed at a pure paleomagnetic study of the Hawaii/Emperor chain. Results were once again conclusive that some shift of either plume location or spin axis played a major role in the formation of the Hawaii/Emperor trend, it was also found that other Pacific plumes had also drifted in a generally southward direction, but not to the same degree as Hawaii.
Tarduno insisted that the hotspot drift was caused by large scale mantle flow deflecting the plume with no component of True Polar Wander. Over time many debates have ensued between supporters of 'fixed plumes with TPW' and the 'drifting plumes with no TPW' models. Tarduno also suggests that TPW over the last 200My may indeed be recording a drift of plume 'groups' in responce to mantle flow (Tarduno focus group summary). It appears that some inconsistencies in both models suggest there elements of both mechanisms are required to explain some observations and sometimes even current understanding and modelling cannot explain some observations (see Iceland Case study)
Tarduno's results from the later, more detailed study of palaeolatitude of the Hawaii/Emperor seamount chain showing the rapid drift of the hotspot from 100 to 60 Ma and the relatively fixed position in the last 40My or so