Particles from class averages outlined in purple and 3D classes colored purple were selected for further processing. Particles that went into grey classes were eliminated from further processing. The final global refinement yielded a map with an overall resolution of 3.07 Å for the core. 3D classification without a mask using particles from the 3.07 Å refinement is shown in Figure 3C (tail containing classes are shown in Figure 1A). To improve map quality at the periphery of the core, multibody refinement was performed for the head, arm, and body lobes, followed by 3D classification of the partially signal-subtracted particles. For the arm lobe, a single good class was identified, which was refined to a resolution of 3.16 Å. For the head and body lobes, two distinct good classes were found, with one containing several extra helices. The classes containing the extra density were selected and refined. The head lobe refined to 3.07 Å, and the body refined to 3.48 Å. During the 3D classification steps classes in purple were used for further processing while grey classes were removed. For 3D refinement steps with three maps the results shown are: (left) refinement without a mask, (center) refinement continued with a mask (mask is shown as an outline around map at its highest threshold), and (right) postprocessed map. For 3D refinement steps with two maps the results shown are: (left) refinement with a mask (mask is shown as an outline around map at its highest threshold) and (right) postprocessed map.