We give an overview of present calculations involving the proton spin structure function. It is shown that a significant part of the discepancy between the data and the Ellis–Jaffe sum-rule may arise through the axial anomaly if the gluons within the proton are strongly polarized. While a quark model, such as the MIT bag, does not include the anomaly, and therefore cannot be expected to reproduce the spin structure function, it does give a rather good description of recent data which is anomaly free, such as the distribution of polarized, valence up-quarks in the proton.