Rocked observatons of the extreme ultraviolet spectrum of the aurora from 550 to 1250 ¿ are presented, and the results for the O I (989 ¿) and O I (1027 ¿) multiplets are examined in detail. The intensity of these emissions and the observed anisotropic radiation field are inconsistent with the relative intensity of the branching radiations. Several possible solutions to the problem are discussed, and evidence is presented that suggests the need to lower the branching ratios by several orders of magnitude.