Nature 429, 853–857 (2004).

It has come to our attention that we failed to cite in our Letter to Nature a key publication1 that describes changes in the magnetocaloric capabilities of Gd5(Si2Ge2) as a result of alloying the compound with small amounts of various metals, including iron. That paper supports the findings presented in our Letter. However, we also show that it is important to consider the magnetic hysteresis in addition to the magnetocaloric effect when assessing the usefulness of a material as a magnetic refrigerant.