On examination of the palaeotemperature succession obtained by Emiliani, particularly in relation to the details of last interglacial isotope stage 5, it is found that the highest palaeotemperature is always recorded early in the stage, and that this peak is followed by a sharp drop in palaeotemperature. It is shown that this palaeotemperature decline is so rapid that even if the climatic optimum of the stage had been climatically suitable for glaciation to commence in North America and Europe, ice could not have accumulated sufficiently fast to change the isotopic composition of the oceans enough to account for the observed change. It is concluded that the change was due to a drop in world temperature. Comparison with the Eemian vegetational succession shows that there is no place for such a substantial temperature decline until the end of the Eemian stage. Thus the Eemian interglacial stage corresponds to only a small part of the isotope stage 5. The age of the Eemian is considered and a value of near 120 000 years (120 ka) is found to fit the available data best. On the basis of this figure, the duration of the Eemian may be estimated from the marine record at about 11 ka. The temperature drop demonstrated comprises a fall to near the lowest glacial level. This temperature decline must have been controlled by a mechanism which did not require the presence of large ice sheets for its effect. The minor glaciation which ensued was likewise terminated by a climatic change which took effect despite the fact that continental ice sheets were by no means fully extended. These conclusions have important bearing on the theory of Pleistocene glaciations.