The origin of the partition coefficient ( D) for Sr in plagioclase is ascribed by BLUNDY and Wood (1991) mainly to the chemistry of the crystal rather than that of the liquid. However, a major data set used to support this conclusion also demonstrates an important correlation of D with the augite content of the liquid. One truth cannot be used to deny another from the same data set. Moreover, it appears that D has a limiting value of 1.0 in the system An-Ab, confirming the importance of liquid composition. The data base of natural magmas said to justify the conclusions of the paper omits the important troctolitic and noritic, low-augite varieties for which the augite effect was specifically claimed, and therefore cannot be used to discount the effect. The data base is skewed toward silicic igneous rocks for which a high value of D is experimentally known. The attempt to rationalize the data from the Kiglapait Intrusion by supposing the escape of low-Sr material ignores the evidence of the chilled margin, which matches the Sr budget of the cumulates, and neglects to mention that the vented mass would have to be equal to that of the entire intrusion to account for the small variation in Sr observed. Much work remains to be done, but it is clear that the liquid plays a much more important role in this partitioning effect than allowed by the authors.