We review solar//geophysical data relating to the great magnetic storm of 14-15 May 1921, with emphasis on observations of the low-latitude visual aurora. From the reports we have gathered for this event, the lowest geomagnetic latitude of definite overhead aurora (coronal form) was /40° and the lowest geomagnetic latitude from which auroras were observed on the poleward horizon in the northern hemisphere was /30°. For comparison, corresponding overhead//low-latitude values of /48°/32° and /41°/20° were reported for the great auroras on 28-29 August and 1-2 September 1859, respectively. However, for the 1921 event, there is a report of aurora from Apia, Samoa, in the southern hemisphere, within /13° of the geomagnetic equator. This report by professional observers appears to be credible, based on the aurora description and timing, but is puzzling because of the discrepancy with the lowest latitude of observation in the northern hemisphere and the great implied auroral height (~2000km, assuming overhead aurora at Auckland, New Zealand). We discuss various possibilities that might account for this observation.